Testing Soviet Economic Policies, 1928-1940
Transcript of Testing Soviet Economic Policies, 1928-1940
FINAL REPORT TONATIONAL COUNCIL FOR SOVIET AND EAST EUROPEAN RESEARCH
TITLE: Testing Soviet Economic Policies,1928-1940
AUTHOR: Holland Hunter andEverett J. Rutan, III
CONTRACTOR: Haverford College, Haverford, PA 19041
PRINCIPAL INVESTIGATOR: Holland. Hunter
COUNCIL CONTRACT NUMBER: 621-6
The work leading to this report was supported in whole or inpart from funds provided by the National Council for Sovietand East European Research.
TESTING SOVIET ECONOMIC POLICIES, 1928-1940
Holland Hunter and Everett J. Rutan, III
Executive Summary, September 1980
The USSR is thought by most of the world to have been transformed from a
backward agricultural country into a powerful industrial state during the
last fifty years. Impatient nationalist leaders in the Third World may
decide that by imitating Soviet methods they can achieve similar results.
But this project provides evidence that Soviet methods—especially the
forced collectivization of agriculture—were in fact tragically ineffi-
cient. The pre-war economic growth potential of the economy was under-
mined through unnecessary policy errors that imposed measurable costs on
the whole economy.
The Soviet regime expected collectivized agriculture to provide a surplus
which could be used to build heavy industry. Simultaneously, farming
could be modernized. Instead, fierce peasant resistance caused a drastic
shrinkage in available animal draft power; investment had to be shifted
away from industry in order to provide tractors, combines and trucks as
replacements. Agricultural output fell, the surplus disappeared and the
whole economy suffered. While these facts have long been understood in
the West, this project provides the first quantitative estimates of the
costs involved.
Our procedure involves fitting early Soviet data—not yet disturbed by
Stalinist pressures—into an input-output (national balances) framework
of ten producing sectors and six categories of final demand. We estab-
lish a baseline for the year 1928, and derive the parameters that
Gosplan apparently perceived as governing the economy's ability to
-2-
expand. The data are used to build a dynamic linear programming model
called KAPROST (from "kapitalnyi rost" or capital growth in Russian).
The solution to this model provides a complete set of national accounts
over six two-year time periods which shows how the economy might have
grown from 1929 through 1940 under "problem-free" conditions. The
economic potential inherent in a growing labor force combined with a
growing stock of fixed capital is thus revealed.
We then modify this problem-free reference solution by altering relevant
parameters to reflect several developments—depression, rearmament,
collectivization—that were not foreseen by Gosplan. The resulting
composite reconstruction produces an expansion path that appears to match
up quite closely with the actual historical record as pieced together by
Western students of the period. The differences between the problem-free
reference solution and the composite reconstruction can be attributed to
specific events and policies, and rough estimates of the size of indivi-
dual impacts can be derived.
The Soviet economy, as modeled by KAPROST in the problem-free solution,
undertakes substantial investment in fixed capital while undergoing
structural transformation and introducing new technology. Production
rises quite rapidly after 1928. By 1940 gross output is almost four
times as large as in the base year, and GNP proper is more than four
times as large as it was in 1928. Agricultural output doubles; industry
and construction produce eight times as much; services output rises by a
factor of 2.7. Moreover, this is accomplished without any belt tight-
ening: consumption never falls below 1928 levels, and by 1940 is over
two and one-half times as large.
These are extremely high rates of growth, perhaps implausibly high. They
reflect optimistic assumptions concerning all the technical parameters
governing production and capital formation, though none of the individual
—3—
parameters lies outside the bounds observable in other times and places.
Japan's record a quarter of a century later lends credence to the possi-
bility of expansion on this scale. It is surely reasonable, then, to
take our reference solution as indicating the upper limits of hypo-
thetically achievable growth.
In adjusting this problem-free reference solution to incorporate actual
developments, we begin by estimating the impact of world depression on
the Soviet economy. Given the extremely low ratio of foreign trade to
GNP in the USSR, it is not surprising to find that the impact was very
modest. The Soviet economy before 1929 had been relatively isolated from
the world economy. After a brief burst of high-technology imports in
1929-32, Soviet authorities spurned further technological transfer. If
pre-depression terms of trade had continued to be available, no doubt
Soviet exports and imports would not have fallen off so severely (though
it would have been very difficult to maintain grain exports after 1932).
History shows, nevertheless, that industrial expansion was not curtailed
by the cutbacks in Soviet imports.
Next we estimated the impact of Soviet rearmament. Toward the end of the
decade direct outlays on national defense became a heavy burden, cutting
sharply into household consumption and civilian fixed capital formation.
However, if one counts military hardware as capital, total capital stocks
are larger under rearmament than in the reference solution. The results
of a solution combining the effects of world trade depression and rear-
nataent are about the same.
The inpact of these two exogenous developments, forced on the USSR from
the outside, was far outweighed by the effects of collectivization.
After further adjusting KAPROST to reflect the actual levels of fixed
capital and output in both the field crops and livestock sectors, we find
that household consumption over the years 1929-1940 is 37% lower than in
-4-
the reference solution. In the worst years it is almost 50% below what
it might have been.
KAPROST also reveals that massive hidden underemployment must have
existed in the Soviet Union in the 1930's. In the model, the same labor
productivities are used in all simulation experiments. Under the pro-
blem-free solution, some of the available labor is not needed in the
early years, when sharp structural shifts are underway. Rearmament
delays full employment until the mid-1930's. But when the effects of
forced agricultural collectivization are added, unused labor rises from
14% to an average of nearly 30% for the decade. Successive constrictions
on the economy make more and more labor superfluous.
In fact, unemployment was not permitted. Soviet authorities kept every-
one at work: on factory payrolls, on collective farms, in the Gulag
archipelago. Average labor productivity in physical terms must have
fallen far enough to absorb the whole labor force (except for the "excess
deaths" which are not included in any of the solutions). Thus massive
unemployment was camouflaged in the USSR, in constrast to the explicitly
exposed unemployment in the West during the same period.
The Soviet drive for economic expansion was also adversely affected by
grim changes in the domestic political environment that are not easily
incorporated in a quantitative model. The purges and mass terror that
peaked in 1937-39 reflected, in part, external threats; but the domestic
tensions induced by forced collectivization of agriculture also seem to
have been a major cause. The purges wreaked havoc in the armed forces,
heavy industry and the railroads as morale was impaired and performance
deteriorated. These destructive influences contributed to the pervasive
way in which collectivization undermined Soviet growth.
-5-
The lesson for other countries is clear. Economic development involves
basic changes in agriculture, but Soviet experience shows how large the
cost of excessive haste can be. Our modeling procedures could be used
not only to examine more recent Soviet developments or to decompose the
past experience of other countries, but also to examine the current
policy choices facing governments. KAPROST is a flexible instrument for
modeling technological progress and structural change, combining a number
of analytic devices already familiar to economic development specialists.
Even those who have no interest in Soviet economic history will thus find
our report and its appendices to be useful.
FINAL REPORT TONATIONAL COUNCIL FOR SOVIET AND EAST EUROPEAN RESEARCH
TITLE:
AUTHORS:
Testing Soviet EconomicPolicies,
HollandEverett
1928 - 1940
Hunter andJ. Rutan, III
CONTRACTOR: Haverford College, Haverford, PA 19041
PRINCIPAL INVESTIGATOR: Holland Hunter
COUNCIL CONTRACT NUMBER: 621 - 6 July 1980
The work leading to this report was supported in whole or in part fromfunds provided by the National Council for Soviet and East European Research,
TABLE OF CONTENTS
Chapter
I Introduction and Summary
II The Economic Development Problem Facing the USSR
III A Computable Model for Analyzing Output Expansionans Structural Change
IV Calibrating the Model to Obtain a Reference Solution
V What Actually Happened
VI Decomposing the Record to Measure the Impact ofMajor Events
Appendices
A Statistical Foundations
B Logic and Algebra of the Model
C Data as Modified for Use by the KAPROST Model
D Programming and Computer Solution of the Model
Pag
1 -
1 -
1 -
1 -
1 -
es
13
12
15
62
26
1 - 6 2
1 - 3 7
1 - 20
1 - 2 5
1 - 3
Comments and criticism welcome.
Address correspondence to:Professor Holland HunterDepartment of EconomicsHaverford CollegeHaverford, PA 19041215-649-9600
Not to be cited without permission
July 1980
Chapter I. INTRODUCTION AND SUMMARY
This study uses new tools in order to dissect Soviet economic experience
during the crucial period from 1928 through 1940 when the foundations of the
present economy were laid. These years saw a massive effort to restructure
the whole economy by means of centralized controls focused on increasing
certain kinds of output at maximum speed. Very substantial output increases
were indeed achieved, and the control mechanism that came into being remains
at the core of the Soviet economy even today.
The purpose here is to obtain a fuller understanding of the steps by
which the Soviet economy was brought from its base-period situation in 1928
to the state that had been reached by the beginning of 1941. Output levels
were raised, some manyfold and others scarcely at all. Input flows were en-
larged and redirected. Sectoral proportions shifted markedly. The trans-
formation occurred very rapidly, under conditions involving great tension and
sacrifice. These were pioneering efforts in rapid economic development before
the concept and goal of systematic economy-wide economic development had spread
across the world. The prewar Soviet record thus has great significance, not
only in its own right, but also for the lessons — both positive and negative—
that it may have for other countries with similar objectives.
The main features of the record are already well established. Sub-
stantial studies, mainly by Western scholars, are already available. The
most fundamental are a set of meticulous analyses reconstructing national
income and product accounts for 1928, 1940, and the intervening year of 1937.
These benchmarks make it possible to compute growth rates over the intervening
periods, using alternative price weights. There are also detailed studies of
individual sectors of the economy, tracing their evolution in detail. Several
economic histories review the broad course of events and several insightful
1-2
analyses evaluate institutional changes. Major studies have investigated
planning theory and planning practice. As a result of all this research, a
fair measure of scholarly consensus exists concerning the broad outlines of
what happened.
Building on this material, it now appears possible to deepen our
understanding of the record by dissecting it and integrating it through
applying an intersectoral, intertemporal model. That is what is done by the
KAPROST (capital growth) model. KAPROST disaggregates the Soviet economy
into ten producing sectors and subdivides the period after 1928 into six two-
year periods. The model makes it possible to trace the direct and indirect
requirements laid on each producing sector in each period by the major forms
of final demand. KAPROST provides reconstructions for each period, not only
of final demand and value added, but also of intersectoral flows. The six
periods are linked together by the process of capital formation, which can be
followed in detail. One can also trace the pattern of structural change in
output and input use, period by period. The model imposes order on relations
among all the economy's parts in each period, while simultaneously enforcing
consistency as the economy moves from one period to the next.
This modeling effort allows the investigator to compare actual develcp-
ments with the economy's potential. Having assembled a consistent summary of
actual developments, we can see if the potential that lay in the Soviet economy
at the end of the 1920s was in fact realized thereafter. Through enlarging
the labor force and reassigning it, through building new capital plant and
equipment embodying advanced technology, and through altering the structure
of output, Soviet authorities were determined to uncover the potential lying
unrealized in the USSR. We attempt to measure this potential and compute a
problem-free reference solution for comparison with actual historical developments.
1-3
Differences between the expansion path traced out by the reference solution
and the sequence of developments that actually occurred reflect the impact
of several major events; our approach permits us to offer crude estimates of
the separate impact of each event. In effect, the actual record is decomposed
so that causal influences are measured.
Initial skepticism concerning the feasibility of estimating the Soviet
economy's potential is certainly in order. One can have doubts about both
the conceptual meaningfulness of a statistical framework capable of answering
the question and also have doubts about the availability of reliable evidence
from Soviet sources. Conceptual doubts can perhaps be removed through detailed
presentation of the model itself; we set this forth in Chapter III. Doubts
concerning the availability of reliable evidence are more easily dealt with.
The key point is that for measuring the economy's potential in the late 1920s
we can rely on ample, detailed Soviet data assembled before Stalinist pressures
distorted the evidence. In fact, Soviet planning agencies had, by 1929,
assembled data for most of the economy in volume and detail that would be the
envy of most developing-nation planning agencies today. As explained briefly
in Chapter II, the planners spent some eight or ten years building up a com-
prehensive data base reaching to all significant aspects of the economy.
Reservations concerning definition and coverage within national aggregates can
generally be removed through inspection of sectoral and regional detail. Though
Soviet standards for statistical explanation fell short of the best Western
practice, the availability of numerous alternative estimates permits an adequate
resolution of most interpretive issues.
Parameters derived from this base-period evidence embody the properties
of the 1928 economy, i.e., labor and capital productivities, input-output
coefficients, worker and peasant consumption patterns, fixed capital gestation
1-4
periods, etc. The first Five-Year Plan, supplemented by related Gosplan and
TsUNKhU material, provides data permitting derivation of the parameters that
can be seen as governing the process by which the economy could expand from
its 1928 starting point along the lines intended by the Party. The parameters
governing expansion are derived incrementally in a way that unhooks them from
the 1932/33 terminal year of the first Five-Year Plan, so that rates of feasible
expansion subject to these constraints can differ from the rates laid down in
the "optimal" version of the first FYP. (It was, in fact, discovery that the
first FYP was infeasible that led to the formulation of the KAPROST model as
an analytic device for uncovering the economy's feasible potential.) In order
to carry the test forward as far as 1940, it is also necessary to supply exogenous
estimates of the labor force, Soviet exports, and government purchases, together
with assumptions regarding a few parameters, all of which can be varied in order
to test their implications. With the addition of an objective function embodying
the purposes of the system's directors, this whole set of relations forms a
linear programming problem that can be easily solved on present-day computers.
Variation of key parameters or variables alters the solution values that make
up an answer. Through systematic experimentation one can feel out the contours
of the economic system constraining the expansion process, discovering also which
parameters and variables have the most decisive impact on the results.
The evidence available on actual Soviet developments comes partly from
official Soviet sources but mainly from Western scholars who have reconstructed
Soviet data in an effort to assure uniform coverage and stable price weights.
From the early 1930s through the mid-1950s, official Soviet data were either
withheld or issued in distorted or misleading forms. Over the last twenty years
or so, retrospective information on the 1928-1940 period has improved, as
official sources have released undistorted series, or as Soviet scholars have
1-5
made use of the underlying evidence available to them, but very little thorough
reconstruction of primary official evidence has been reported. In spite of
large remaining gaps, it has proved possible to assemble a record adequate to
the needs of this project.
The KAPROST model has several distinctive features, designed to make it
match up with the special character of the Soviet system. Like all linear
programming models it is purposive rather than passive, in the sense that its
activities are directed toward maximizing the value of an objective function.
In development modeling the maximand is usually household consumption, which
appears in the KAPROST objective function as well. Maximization of household
consumption over several periods requires that current resources go into fixed
capital formation in order to enlarge the capital stocks of sectors delivering
final demand to households. But since gross output is delivered to other forms
of final demand as well: public consumption, national defense, and non-defense
government final use, some capital formation is motivated by non-consumer
objectives. In the USSR the system's directors have seemed to attach independent
value to growing stocks of fixed capital as symbols of strength. In Marxian
terms, accumulation has at least equal weight with consumption as a goal of
economic activity. The KAPROST objective function therefore includes terms for
stocks of fixed capital in each of the ten sectors at the end of 1940, and the
weights on consumption terms and capital stock terms can be varied to reflect
alternative policy emphases.
The model includes both demand-side and supply-side constraints on out-
put expansion, but the relationships involving effective demand and the dis-
position of income are secondary, while the real, physical limitations on
activity levels and additions to capacity are the decisive constraints. In
this respect KAPROST accurately matches the Soviet system of the 1930s, when
1-6
through forced saving and taxation the authorities overrode any private un-
willingness to finance nonconsumption activities. In the same vein the model
provides flexible levers for manipulating the flows of output going separately
to workers and peasants so that the extent of belt-tightening can vary from
period to period for each part of the population. It is thus possible to
recreate the changes that occurred and compare them with hypothetical alterna-
tives .
Perhaps the most distinctive features of the model relate to the way
technological progress is brought into play. Each producing sector can draw
on two stocks of fixed capital plant and equipment. One stock is the old
capital on hand at the end of 1928, embodying prevailing technology. The
other stock existed at that time only as small amounts of unfinished new capital,
to be augmented thereafter through investment. This new capital embodies
advanced technology, in the literal sense that it incorporates the best practice
available from the West at the end of the 1920s. The old capital in each sector
simply depreciates, while new capital comes into use — after a specified
gestation period — in those sectors toward which investment is directed, at a
speed determined by the optimizing procedure. As advanced-technology capital
replaces old-technology capital in each sector, technological progress becomes
visible, in a pattern and at a rate of speed reflecting the purposes of the
regime interacting with the potentials in the situation. In addition to this
flexible procedure for modeling the transition to use of improved technology,
we add a parameter that reflects the fact that new capital does not produce at
100% of capacity until it has been "mastered."
This vintage approach to technological change is peculiarly appropriate
for Soviet experience in the 1930s because after an initial burst of imports
from Germany, the US, the UK, France, and a few other suppliers of machinery
1-7
and equipment, the USSR devoted the rest of the decade to "mastering technique",
importing very little and making only modest changes in the technology embodied
in their rapidly-growing stock of fixed capital.
In choosing an optimal degree of disaggregation for this project, we
were conscious of the desirability of reflecting major policy emphases and
alternatives while simultaneously holding down data requirements and preserving
interpretability. Classical input-output models often involve several dozen
sectors, but in a multi-period programming model this degree of disaggregation
quickly becomes unmanageable. On the other hand there was no need to be confined
to a two-sector model like that employed by Gisser and Jonas in their paper on
"Soviet Growth in Absence of Centralized Planning". Our concern for capital
formation suggested that a construction sector should play a distinct role,
and our stress on changes in the structure of capital suggested that sectors
with large stocks of fixed capital in 1923 should be recognized, perhaps even
subdivided.
For these reasons we have rearranged the underlying evidence so as to
form ten producing sectors. Agriculture is subdivided into field crops
(together with forestry) and livestock products (together with fishing and
hunting). Industry is subdivided into producer goods and consumer goods.
Production of electric power, at the very center of Party priorities, is kept
separate from producer-goods industry to form its own sector, and construction
is a separate sector as well, playing a key role. Transportation and
communications make up a large, capital-intensive sector, and housing (both
urban and rural), though given low priority by the Party, makes up another
distinct sector with a large stock of inherited capital. Finally, though
considered non-productive in Marxian terms, a trade-and-distribution sector
and a government services sector are recognized as significant users of labor
I-8
and capital. Though interindustry flows among branches of industry are not
traceable in any detail under this approach, our sectoral scheme suffices to
illuminate the key policy issues of the period.
To set the stage for a detailed reconstruction of the base period
situation, we begin in Chapter II with a brief summary of Soviet economic
goals, the resources available and two special factors that conditioned
Soviet efforts. These features of the Soviet situation, in turn, have influenced
the structure of the model we have designed to analyze the Soviet record.
First, as to goals, the fundamental economic aim was "to catch up with
and surpass the advanced countries." This meant primarily a drive to build non-
argicultural activities in order to achieve three sub-objectives. Modern
industry would enable the regime to win its contest with the peasantry through
supplying them with tractors, machinery, and the modern methods that would
wean them away from their petty bourgois ways. Building a strong heavy indus-
trial base would defend the country against hostile capitalist encirclement.
In long run terms, a modernized economy would generate high labor productivity
and living standards, thus demonstrating the superiority of the Soviet system.
In working toward these goals, the USSR could make use of an ample labor
force and a rich base of land and natural resources. There was also an im-
pressive stock of fixed capital already in existence, but, as shown below in
Table III-l, Soviet fixed capital in 1923 was chiefly in agriculture, transpor-
tation, and residential housing. Industry accounted for only 9% of the total,
suggesting that the drive to catch up with and surpass should be concentrated
on building industrial capital.
In seeking to formulate and carry out economic development plans, the
authorities could draw on almost a decade of data collection and
1-9
experience. Statistics with steadily expanding coverage provided base-period
evidence that could be used for projecting changes. Both government departments
and producing enterprises had gained some experience with efforts at operational
planning.
Two special factors conditioned interaction between these goals and the
wherewithal applied to them. One was the country's international political
situation. Moscow saw itself surrounded by hostile capitalist powers unwilling
to provide economic support except on the harshest commercial terms. The
Soviet authorities sought economic independence, even if it had to be achieved
through initial dependence on technology transfer from the West. Thus by
comparison with the practice of most developing nations today, Soviet plans
have from the beginning given only a small role to foreign trade.
A second conditioning factor was the Party's attitude toward prices and
markets. Since prices and markets were seen as the operational essence of
capitalism, Party members from top to bottom were suspicious of then. The
Party was predisposed instead toward using administrative, non-market instru-
ments for guiding economic activity. On ideological grounds, physical alloca-
tion of supplies, together with administrative allocation of budget funds, were
strongly preferred over responses to market demands.
Our model therefore provides much less detail on the economy's external
relations than is usually found in economic development models, though the
specification does permit some crude testing of the impact of world depression
on the Soviet economy. The model also stresses relationships that, conceptually
at least, run in real terms. Here, too, the emphasis is rather different from
that of the usual development model. In linear programming and input-output
economics, the real side and the value side are inextricably bound toghther,
but in moving the levers that determine model outcomes , we will be evaluating
I-1C
Soviet experience in manipulating the "real" levers.
In chapters III and IV we present the statistical evidence underlying
our effort to measure the economy's potential at the end of the 1920s and
explain how the KAPROST model was calibrated to match the data. We
ask: how rapidly could output be expanded, in the directions intended by
the Party, under problem-free conditions? Marked structural change in the
composition of output and inputs was sought, subject to the constraints
imposed by the existing situation and the time required to bring about
changes. We extract specific measures of the constraints through incremental
dissection of the data prepared by Gosplan and TsUNKhU for the first Five-
Year Plan, unhooking the parameters from the year 1932-33, when the first FYP
was supposed Co end, and making them part of a flexible analytic framework.
On the basis of this evidence, KAPROST computes a problem-free solution
showing in ten-sector detail how the Soviet economy might have expanded from
1928 through 1940 if planners' expectations had prevailed. In early 1929
they did not anticipate the collectivization of agriculture that swept across
the country in Che fall and winter of 1929-30. Nor d i d they anticipate the
world depression that turned the terms of trade sharply against Soviet
commodity exports and imports. Finally, chey made no provision for channeling
current output into defense procurement. We therefore obtain a reference
solution embodying the optimistic hopes of the system's directors in the absence
of these three negative developments, each of which made expansion more
difficult. In several respects the reference solution is perhaps unduly
optimistic, though it bears no resemblence to the widly unrealistic targets that
were bandied about in the early 1930s. It illustrates the upper range of
outcomes reflecting the economy's potential.
1-11
The central message of our reference solution is that the Soviet
economy in the 1928-1940 period contained a very substantial potential for
output expansion, under problem-free conditions, without much need for belt-
tightening. In view of the situation that prevailed in the base period, no
elaborate model is needed to suggest such a conclusion, since the resources
and growth possibilities were obviously at hand. The model provides systematic
and consistent estimates, however, of the gains that might have been achieved.
The Soviet gross national product, for example, might have risen more than four-
fold, and the stock of fixed capital at the beginning of 1941 might have been
five times larger that it was at the beginning of 1929. Consumption would have
increased some 2.7 fold while investment was rising from an index of 100 to
an index of 824.
Chapter V presents the best available current evidence on actual economic
developments, in continuous annual time series, over the 1928-1940 period, for
our ten sectors. The evidence is compiled mainly from Western sources, free of
the distortions that afflicted official Soviet figures. An appreciable margin
of error remains, no doubt, but the series appear to provide a workable basis
for appraising the course of events. These statistics of course reflect the
impact of agricultural collectivization, world depression, rearmament, and all
the other difficulties that struck the Societ economy. Can they be built into
KAPROST?
In chapter VI we show that they can. We begin by altering the export
and import parameters of the reference solution to incorporate the impact of
world depression on the domestic Soviet economy. It proves to be slight. We
then introduce Soviet rearmament into the model, taking account both of defense
procurement and of other claims by the armed forces on the economy. The defense
I -12
impact is appreciable, especially toward the end of the decade. We also test
the combined influence of world depression and rearmament, seeking to show the
joint influence of these two external forces on the Soviet system. 3y comparison
with the problem-free reference solution, we find that though total gross outputA
is not much affected, household consumption is reduced about 22% in the mid-
Thirties and about 43% at the and of the decade.
The greatest damage is caused, however, by agricultural collectivization.
Adjusting KAPROST to reflect collectivization lowers gross output substantially
throughout the whole period, while imposing a drastic cut in household con-
sumption, which is 32% lower in the mid-Thirties and 55% lower over the last
four years. The loss of draft animals and productive livestock due to collect-
ivization did far more than reduce the flow of livestock products to light
industry and to the population. It forced industry to deliver tractors and other
agricultural machinery to field-crops agriculture more promptly than had been
intended, thus making non-agricultural capital formation more difficult. It
also lowered labor productivity, both directly in agriculture and indirectly
throughout the economy, Probably the most important consequence of collec-
tivization, though it is not directly measurable by KAPROST, was the political
tension and pressure that this "second revolution" spread throughout Soviet
society, culminating in political purges and mass terror. This is the tragic
human story that lies behind the laconic figures in our computations.
Rearmament and collectivization, especially collectivization, also
prevent the economy from making effective use of the whole labor force. In
our problem-free reference solution, after structural shifts have been accom-
plished, KAPROST employs all workers and peasants from 1933 through 1940.
In our composite reconstruction, reflecting the impact of collectivization and
rearmament, somewhere between a fifth and a third of the labor force is super-
1-13
fluous, at least given the labor productivity that prevailed in 1928 and was
expected for the 1930s. The unemployment shown by KAPROST was in actuality
camouflaged by government policy. The Constitution of 1935 declared that
unemployment did not exist in the USSR, and in fact everyone was kept at work,
on payrolls or in kolkhozy or in corrective labor camps. Average labor produc-
tivity per person in physical terms fell far enough below its potential to absorb
the superfluous labor. Massive unemployment was thus concealed, by contrast
with the explicitly exposed unemployment in the West during the same period.
To summarize this exercise in ex post planning, we suggest that its
findings be taken with a grain of salt. The statistical evidence is fragile
and the model is rather crude. With better data and a more sophisticated
model, the results could be made more precise. It is not likely, however, that
their implications would be significantly altered. This period in Soviet
experience was a harsh one. Great sacrifices were made, whether necessary or
not, and much was accomplished, however wastefully. Let us hope that other
countries will be able to carry out these tasks at lower costs.
IV-1
Chapter IV. Calibrating the Model to Obtain a Problem-free solution
In this chapter we report on computations intended to uncover the
potentials that lay in the Soviet economy at the end of the 1920s, had it
expanded along the lines intended by the Party and spelled out in the first
FYP. We employ the data derived from the first plan and several related
Gosplan documents as described in the preceding chapter. These base-period
levels and associated parameters are fitted to the KAPROST model, which then
computes a complete set of values for output, employment, capital stocks,
etc., over the next six two-year periods through the end of 1940. After a
series of adjustments we obtain a problem-free solution which can then be
compared with actual historical developments in an effort to trace the impact
of major events on the economy.
The process of establishing a good reference solution itself requires
a series of adjustments. We found it necessary to restate the specification
of both our household consumption equations and our formulation of the post-
terminal constraints on capital formation and to modify somewhat our base-
period fixed capital estimates. After describing these calibration adjust-
ments, we present a reference solution, one designed to reflect a balanced
emphasis on both capital accumulation and household consumption. This base-
line solution is not meant to be optimal in a welfare sense, since we lack
the evidence required to form such a judgment. The baseline solution illus-
trates, however, the general dimensions of the expansion that might have un-
folded over the 1928-1940 period in the absence of any complications.
IV-2
Evolution of the Consumption Equations
The treatment of consumption and its structure in KAPROST is fairly
straightforward. However, experience forced us to modify an initially rigid
structure in order to obtain economically reasonable solutions. In this section
we describe the motivation for and precise forms of those changes.
Table IV-1 reproduces the model's consumption-related equations. Income
of both peasants and workers is determined by wage payments to those who are em-
ployed (equations 1.1 and 1.2). We relate total consumption to income by means
of an "average propensity to consume" parameter— <X for peasants, 23 for
workers—in equations 2.1 and 2.2. These parameters may also be viewed in
another way. Any KAPROST solution which emphasizes investment over consumption
will cause the constraints in equations 2.1 and 2.2 to be binding, that is
equalities. As we do not provide for private savings, (1- c<) and (1-/3 ) may be
considered inflation or taxation rates: the factors by which the government re-
duces the real value of wages paid. It is entirely consistent with the Soviet
record for the model to provide a parameter by which consumption may be lowered
in order to divert output to other uses.
Total consumption must be broken down into demands for the output of
specific producing sectors of the economy. Equation 3 was our initial speci-
fication. The parameters '"f^ and uJ were vectors representing how one ruble of
peasant and worker consumption (respectively) was allocated among the outputs
of the producing sectors. The shares were derived from actual 1928 and planned
1933 consumption figures (as described in Appendix C).
When the model was solved using equation 3 (with oC and /2> set at .7,
close to historically observed values) we obtained solutions which left large
amounts of unused capital and unused labor in all periods. This is totally at
variance with the full employment, full resource utilization policy of the
IV-3
Soviet Union. The computed outcome reflected a lack of capacity in two
sectors: housing and trade-and-distribution, both of which produce output
primarily for consumers. As worker and peasant income rose, the fixed pattern
of consumption placed rising demands on these sectors, and when capacity was
exhausted, no further employment was possible. This bottleneck could be
alleviated by lowering c< and p> to .5 or at times .3, thereby spreading the
output of the consumer - critical sectors more thinly. However, there was no
means for the model to increase worker and peasant consumption in other sectors
to compensate.
The Soviet (and most developing country) experience is that housing, trade
and distribution, civilian transport and communication—in sum the entire con-
sumer infrastructures—are allowed to be overcrowded and overutilized, even to
deteriorate, in the initial stages of growth. Consumers are at least partially
compensated by increased availability of food and manufactured items. Hence the
image of the Soviet worker who waits in lines to buy everything, and then takes
it back to his—by our standards—overcrowded apartment.
In modifying KAPROST, therefore, we needed a way to limit the portion of
consumption which was required to be satisfied in strict proportions. In some
sense these proportions represent consumer preference, as Soviet markets in
1928 were relatively free. (A second reason to retain their influence in the
model was to insure that peasants and workers received a minimal selection
from all sectors producing consumable goods.) But we also wanted to introduce
the capability for the solution to provide the unrestricted portion of total
consumption from any sector with slack capacity. Clearly this would have to
be examined for reasonableness in each solution; it would not make sense, for
example, to provide lopsided amounts of producers goods for consumption. These
two provisions, limiting the sector-restricted portion of consumption, and
IV-4
providing compensating consumables from other sectors, along with our ability
to control the real value of income in terms of total consumption, would give
us the flexibility appropriate to the situation we were modeling.
Equations 4.1 and 4.2 were finally used in KAPROST, reflecting these
desiderata. In 4.1, the consumption demand on any producing sector consists of
portion,p , of total peasant consumption CP multiplied by the consumption
structure vector "57"" ; a similar contribution for workers with V , CW and CJ ,
respectively; and an amount of optional consumption, OC. Equation 4.2 restricts
the sum of the fixed-proportion components of consumption ( y°, CP and § CW) plus
all optional consumption, to equal the total of worker and peasant consumption.
With this structure we could obtain model solutions which combined full-employment
with reasonable levels of consumption. By paying attention to capacity, and ad-
justing the oC 9 A , jQ and ~? parameters, we are able to provide the contribution
to optional consumption realistically from sectors such as agriculture, consumer
goods, etc. The reader will note that the model cannot allocate optional con-
sumption to workers and peasants directly, as two sets of OC variables would not
be independent.
IV-5
Table IV-1
Evolution of the Consumption Sector
YP = peasant incomeYW = worker incomew = wageL° = labor employed, old technology industryLn = labor employed, new technology industryCP = total peasant consumptionCW = total worker consumptionF = consumption demand to a particular sectorOC = optional consumption
IV-6
Treatment of Sectoral Proportions Among Terminal-period Capital Stocks
As pointed out earlier, the Party intended marked structural change
in the sectoral proportions of Soviet fixed capital stocks, change that
would even be evident by the end of the first plan period. Longer-term
analyses foresaw further structural change, but without specifying any de-
tailed sectoral estimates. Some mid-point evidence is provided by targets
for 1937 in the second FYP as confirmed in January 1934. In Table IV-2 we
show the percentage shares of each of our ten sectors in total fixed capital
stocks in 1928, as intended for 1938, and as projected to the beginning of
1941, assuming constant absolute growth and assuming constant percentage growth
at 1928-1933 rates. The patterns differ quite markedly and no clear inferences
can be drawn from them. The 1933 and 1941 distributions reflect underlying
absolute estimates in 1928 prices, while the distribution for 1 January 1938
reflects computations in the second FYP, using so-called 1933 prices whose
underlying basis is not well understood. In projecting forward the 1928-1933
trends, we have deducted depreciated old-technology capital from the terminal
figure in order to focus on investment intentions. One notes that small but
rapidly growing sectors, like electric power and construction, increase their
shares markedly under a constant percentage growth assumption, while sectors
with large ini t ial shares, like housing and transportation, retain large shares
under projection using constant absolute increments.
It would be technically feasible to impose any one of these patterns
on the model, requiring i t to bring into being terminal-year fixed capital stocks
in precisely these proportions. Apart from the uncertainty of choosing among
these alternative muddy clues to Party intentions, a further consideration
militates against imposing rigid constraints here. The Party attached high
priority to electric power and producer-goods industry, and much lower priority
to housing and agriculture. But surely as the economy grew in these directions,
IV-7
emerging opportunities would have been seized upon so that outcomes within a
reasonably wide range of variation would have been acceptable. Our modeling
procedures permit us to juggle weights to simulate a variety of computed out-
comes, choosing those with the most attractive composite results. This is
far preferable to imposing rigid proportions that would force the optimizing
procedure to meet specifications requiring needless sacrifices of output.
Table IV-2. Structure of Fixed Capital Stocks, By Sector, Actual 1928,Intended 1938, and Intended 1941, in percent shares of thenational total.
1928 1938 1941(a) 1941(b)
Agri., field crops 13.2 8.0
Agri., livestock products 11.3 3.6
Electric power 1.0 4.8
Industry-producer goods 4.7 25.0
Industry-consumer goods 4.3 8.7
Transport & communications 18.1 20.7
Construction 1.0 2.0
Housing 36.6 13.7
Trade & distribution 0.8 1.2
Government services 9.0 12.3
(a) Assuming constant percentage growth at 1928-1933 rates*(b) Assuming that the intended 1928-1933 absolute annual increments continued.
Source: 1928 and 1941 estimates derived from computations for active fixedstocks at the beginning of the year. Depreciated old capital isdeducted from the projected 1941 stocks. The 1938 estimates arebased on absolute data in an English-language edition of the secondFYP, p. 265. Agricultural capital subdivided in proportion togross output; electric power capital derived from investment dataand deducted from Group A industry; construction capital set equalto the figure for "miscellaneous" in the Gosplan table.
6.7
4 . 3
16.4
23.2
5.3
11.4
5.6
9.5
9.9
7.7
11.9
8 .1
5.5
15.0
6.5
17.0
3.4
18.4
3.9
10.3
IV- 8
Evolution of the Post-Terminal Constraint
Economists who have used dynamic optimizing models recognize the importance
of properly specifying terminal conditions. Unconstrained, the "optimizing devil"
will "eat, drink and make merry" in the end, making no provision for a tomorrow
which it does not know exists. In a growth model such as ours, this behavior
takes the form of a failure to invest during the terminal, and perhaps earlier,
periods.
The task of the terminal constraint in KAPROST is, therefore, to maintain
investment at economically reasonable levels even whan the resulting capital
stock will not improve the value of the objective function. If these reasonable
levels could be pre-specified, the form of the constraint would be very simple.
At a minimum, this suggests that it would be plausible to provide for enough
investment in new capacity to continue at the rate already achieved. Hence
the stringency of the terminal constraint will depend on the optimal solution
values, and its form must be built around them.
Table IV-3 lists the various forms of post-terminal constraints experimented
with at various times in KAPROST. The last equation is the one currently in
use. The naming conventions used are described fully in appendix B; the footnote
to the table is only intended as a reminder. Note that common to each form is a
post-terminal growth parameter. Q which allows us to tighten the constraints
parametrically. Further, all are in terms of the increment to capital stock in
the second post-terminal period. As the investment structure in KAPROST requires
two periods for building capital, this increment draws on fixed investment in the
terminal period. (The increment in period T+l contributes to the objective
function, so need not be maintained by constraint.)
Equation 1, the initial version of the constraint, requires that investment
be maintained at a level to cover total depreciation (plus whatever "extra" is
required by the Q parameter). This was felt to be unsatisfactorily low. Given
IV- 9.
Table IV-3
KAPROST Post-Terminal Constraints
increment in new technology capital stock,
for sector i, period t
new technology capital stock, sector i.
period t
as above, for old technology.
depreciation rate, new technology capital, sector i.
depreciation rate, old technology capital
post-terminal growth parameter.
IV-10
a high growth solution, with most of the increment to capital stock in the
final period, the increment provided for capital in period T+2 would be
significantly less than the increment in the previous period. While increasing
X could obviate this somewhat, one can't know how high to set o unless one
knows the solution values. However, altering Q changes the solution, and so
on. It is worth emphasizing that a parameter like 0 is meant to allow the
model user to tighter the constraint, and not to correct flaws in specification.
The second equation calls for the T+2 capital increment to cover the de-
preciation of new capital, and to provide for the same increase from period
T+l to T+2 as the solution calls for from period T to T+l. This is a "same
growth" requirement (with O ^ O indicating somewhat more than same growth),
seemingly very reasonable. However, the optimizer interpreted this quite
differently from the way we intended. Our solutions showed this constraint
raised the cost of increments to capital stock in period T+l so much that it
was optimal to build a very large stock by period T, and then allow it to
depreciate. Hence rather than maintaining investment in the last period, we
had inadvertently eliminated it in the last two periods!
Equation 3 gives our final version of the terminal constraint. The capital
increment in period T+2 must equal the average increment in the preceding six
periods, plus some fraction 0 of capital stock in T+l. The Q parameter is
set equal, initially, to the level of depreciation in each sector. Hence the
constraint requires the model to cover depreciation and to maintain the average
per-period increment in fixed capital. In practice this specification has
worked well. However, one might suggest a different pattern for the weights on
previous capital increments, instead of a simple average. We considered this,
until we realized that any altered set of weights would force investment arti-
ficially away from the periods with the large weights as these periods would be
contributing more heavily to the post-terminal burden. As we had succeeded in
IV-11
maintaining investment expenditure, we felt that the work involved in a more
sophisticated weighting scheme was not justified.
Adjustment of Initial Capital Stock and Capital Productivity
In KAPROST two periods or four years are required to add to capital stock.
This implies that maximum output and capital stock levels for the first two
periods are fixed by the choice of exogenous values for initial capital stock,
initial increments to capital stock, and output-capital ratios. Our initial
experience with values calculated from Soviet data (see Appendix C) revealed
two problems: in most sectors capital stock and productivity ratios would not
allow the model to produce output at historic levels; initial capital stock
and increments would not permit the model to attain historic capital stock
levels in the second period. This indicated shortcomings in the data, especially
the estimate of initial increments to capital, the projects in progress in 1928
before the start of the simulation.
Altering the output-capital ratios for the first two periods to allow the
model to achieve historic production levels is not a major problem. These
coefficients tend to fluctuate in any economy, and we know that in the early
1930's various productivity campaigns caused a sharp rise in output above
expectations. For the remaining four periods the output-capital ratios can be
reduced back to anticipated long-run values.
Altering initial capital stock levels, however, is, in essence, giving
the model extra resources at no cost. Higher levels of output are paid for
by direct demands on production through the input-output matrix, and indirect
demands on production through employment and consumption. Higher initial capital
stocks provide free increments in capacity throughout the model's simulation
horizon, diminished only gradually by depreciation. Clearly this is not acceptable.
We resolved the capital stock problem by increasing the initial increments
IV-12
to capital stock (variable 3 ) , the initial investment pipeline, and by
changing the constraint in which it appears from an equality to an inequality.
This is appropriate for three reasons. First, the increments, if large enough,
will allow the model to achieve historically observed capital stocks for the
second simulation period, 1931-1932. Second, by increasing the initial incre-
ment, rather than the initial stock, of capital, the added capacity can be
obtained only at a price. In terms of the investment dynamics in KAPROST, the
initial increment is in its second investment period (third and fourth year
of construction) in the first simulation period (years 1929-19°'")). As three-
fourths of the cost of investment is incurred in the second investment period,
the model solution must use three rubles of current output for every four rubles
of capital it chooses to build. Taken together with the fragile underlying
evidence on the volume of development projects in progress when the model
solution starts, this formulation appears quite plausible. Third, by requiring
the model to add at most the exogenously specified increment, rather than
exactly the increment, we allow some of these pre-plan projects to be left
forever unfinished. This allows the solution more of a choice in terms of
investment versus consumption in the first period. In actual practice, the
optimizing procedure always chose to invest in the increment, and would have
built even more capital if allowed.
Our precise adjustments are indicated in Tables IV-4 to IV-7. Table IV-4-
shows indices of fixed capital stock by sector on Jan. 1 of the years 1928,
1929 and 1931, with 1928 as a base year. In Table IV-5 we calculate the value
of the initial increment in capital stock needed to permit us to match the
historic levels implied by the indices in Table IV-4. To do this we calculate
the initial stocks of old and new technology capital, and subtract the sum
from the historic stock. In any sector where the required increment exceeds
IV-13
Table IV-4. Actual Fixed Capital Stock, January 1, in millions of rubles
Sector
- iAL
EP
PGi
CG
TC
CO
HO
TD
GS
1928Stock |
8 224
7 594
600
3 017
2 618
11 366
595
22 982
487
5 680
Index
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
1929Stock Index
8 553
7 898
676
3 399
2 570
11 583
670
23 478
508
5 920
104.00
104.00
112.67
112.67
98.17
101.91
112.67 ;
102.16
104.23
104.23
1931Stock
9 889
9 132
1 114
5 603
2 508
12 886
1 105
24 258
570
6 652
Index
120.25
120.25
185.72
185.72
95.79
113.37
185.72
105.55
117.12
117.12
Source: Compiled from primary Soviet sources and adjusted as explained
in Appendix A, pp. 23-27. Some of the figures here differ slightly
from those in Table 7 of Appendix A, which reflect recent revisions
and adjustments. Though recomputation of the reference solution
has not yet been carried through, it is not likely to alter
results appreciably.
Sector
AF
AL
EP
?G
CG
TC
CO
HO
TD
GS
(1)
0K1
7 509
701l
569
2 861
2 485
10 959
536
22 086
463
5 519
(2)
0K2
6 259
5 975
511
2 573
2 238
10 188
435
20 397
419
5 210
(3)
NK
1
9 79
954
248
524
452
649
131
1 404
164
246
(4)
Depr.NK1
837
814
223
478
411
603
109
1 305
148
233
(5)
(2)+(4)
7 096
6 789
734
3 051
2 649
10 791
544
21 702
567
5 443
(6)
K1931
9 889
9 132
1 114
5 603
2 508
12 886
1 105
24 258
570
6 652
(7)
AdjustedN32
2 793
2 343
380
2 552
- 141
2 095
561
2 556
3
1 209
(8)
OriginalN"22
315
1 140
194
1 308
697
331
159
479
74
213
(9)ReferenceSolution
N22
2 793
2 343
380
2 552
697
2 095
561
2 556
74
1 209
(1) Initial Stock of old technology capital, Appendix C, Table 6.(2) Column 1, depreciated by sector.
(3) Initial stock of new technology capital, Appendix C. Table 6.(4) Column 3, depreciated by sector.(5) - (2) + (4).(6) Historic 1931 capital stock, Table IV-4 above.(7) m (6) - (5), adjusted increments to capital in period 2.(8) Initial increments to capital in period 2 as originally calculated, Appendix C. Table 5.(9) - max ((7), (8)), increment to capital in period 2 used in reference solution.
Table IV-5. Adjusted Initial Increment to Capital Stock,
AF
AL
EP
PG
CG
TC
CO
HO
TD
GS*
3.15996
2.15026
1.35126
4.26434
5.53981
0.35464
13.07353
0.24250
3.87492
0.90313
979
954
248
524
452
649
131
1 404
164
246
3 094
2 051
335
2 235
2 504
230
1 713
340
635
222
22 752
10 320
1 649
42 032
15 189
5 878
13 068
3 606
6 043
7 710
19 658
8 269
1 314
39 797
12 685
5 648
11 355
3 266
5 408
7 488
7 509
7 01l
569
2 861
2 485
10 959
536
22 086
463
5 519
2.61793
1.17944
2.30931
13.91017
5.10463
0.51538
21.18470
0.14788
11.68035
1.35677
2.67558
1.62549
1.88666
9.72754
5.46371
0.36284
13.06891
0.14742
11.60164
1.04014
2.67558
1.62549
2.30931
13.91017
5.46371
0.51538
21.18470
0.14788
11.68035
1.35677
(1) New technology output-capital ratio, Appendix C, Table 17.(2) New technology capital in period 1, Appendix C, Table 6.(3) = (1) x (2) - New technology output in period one.(4) Historic output in period 1.(5) = (4) - (3) = Old technology output in period 1.(6) Old technology capital stock in period 1, Appendix C, Table 6.(7) = (5) ÷ 6 = adjusted old technology output capital ratio in period 1.(8) Originally calculated old technology output-capital ratio in period 1, Appendix C, Table 17.(9) • max ((7), (8)) - Old technology output-capital ratio used in reference solution periods 1 and 2.
* Government Services output levels are the linearly interpolated values from Appendix C, Table 3.
Table IV-6. Adjusting Old Technology Output-Capital Ratios for First Two Periods.
(1) (2) (3) (4) (5) (6) (7) (8) (9)
Adjusted Original ReferenceN N 0 0 0 0 00
b K X X K b b b , bSector 1 1 1 1 1 1_ 1 1 1 2
Table IV-7. Adjusted New Technology Output-Capital for Second Period
(1) (2) (3) (A) (5) (6) (7)
Sector
AF
AL
EP
PG
GG
TC
CO
HO
TD
GS*
X
0
216
7509710
1179
35787
12230
5251
9211
3016
4894
6196
X2
23599
7567
2738
53632
15512
8615
20587
3717
5429
9279
X
N
268490
1559
17845
3282
3364
11376701
534
3083
AvailableN
K2
2 792
2 454
489
2 264
899
2 070
502
3 094
200
1 079
AdjustedNb2
2.45308
—
3.18814
7.88207
3.65072
1.62512
22.66135
0.22657
2.67000
2.85728
OriginalNb2
3.15996
2.15026
1.35126
4.26434
5.53981
0.35464
13.07353
0.24250
3.87492
0.90313
ReferenceN
b2
3.15996
2.15026
3.18814
7.88207
5.53981
1.62512
22.66135
2.24250
3.87492
2.85728
(1) Old technology output in period two, obtained by depreciating values in Table IV-6, Column 5 above,(2) Historic output in period two.(3) = (2) - (1) = New technology output in period 2.(4) Available new technology capital stock in perios 2, obtained by depreciating values in Table IV-6,
Column 2 above, and adding in 70 percent of Table IV-5, Column 9 above (to allow for slowmastering of newly produced capital).
(5) = (3) • (4) = Adjusted new technology output-capital ratios for period two.(6) Originally calculated new technology output-capital ratios for period 2, Appendix C, Table 17.(7) = max ((5), (6)) - new technology output-capital ratios for second period used in reference
solutions.
* Government Services output levels are the linearly interpolated values from Appendix C, Table 3.
IV-17
Table IV-9. Output-Capital Ratios Used in Reference Solution,(By Period and Sector)
Old Technology New Technology
Sector
AF
AL
EP
PG
CG
TC
CO
HO
TD
GS
Source:
Period1.2
2.67558
1.62549
2.30931
13.91017
5.46371
0.51538
21.18470
0.14788
11.68035
1.35677
Tables IV-6.
Period3-6
2.67558
1.62549
1.88666
9.72754
5.46371
0.36284
13.06891
0.14742
11.60164
1.04014
IV-7. and Append
Period1.3-6
3.15996
2.15026
1.35126
4.26434
5.53981
0.35464
13.07353
0.24250
3.87492
0.90313
ix C, Table 17.
Period2
3.15996
2.15026
3.18814
7.88207
5.53981
1.62512
22.66135
0.24250
3.87492
2.85728
our earlier estimate for the initial increment, we insert the required incre-
ment in place of the original estimate.
In Tables IV-6 and IV-7, we calculate the adjusted output-capital ratios,
using the adjusted initial increment to capital stock. We increase the pro-
ductivity of old technology capital to whatever level is required to achieve
historic levels of output in the first solution period (1929-1930), after
allowing for the contribution of new technology capital. This productivity is
assumed to continue in the second model solution period (1931-1932), and we
increase the productivity of new technology capital stock in that period to
cover any shortfall. Note that we take into account the mastering process
which makes the additions to capital stock in any period only 70% effective
IV-18
during its first two years. Note further that we never decrease output-
capital ratios below the values calculated in Appendix C. Finally, Table
IV-8 shows the output-capital ratios that are employed for the reference
solution. Note that adjustments occur only in periods one and two.
Desired Characteristics of the Reference Solution
The logical first step in developing a reference solution is to take
the adjusted equations and data, set reasonable values for the consumption
and income parameters, set all objective-function weights at 1.0, relative
to the final period (hence revising upward the weight on consumption in
earlier periods by an appropriate power of the social discount factor),
and then examine the result of the optimization algorithm. It should be
obvious, however, that we still have a great deal of freedom in determining
the characteristics of the solution in these last chosen parameters: average
propensity to consume, fraction of consumption obtained in desired propor-
tions, and objective function weights. Hence the first step, really, is to
attempt to define what properties a reference solution ought to have, and
then see if they can be achieved by varying these control parameters.
Note that we are not imposing a solution on the model; we are no longer
changing any given economic data. Rather we are selecting one out of an
infinite range of solutions inherent in the data. By varying, as Soviet
planners might, the real value of income, the proportion of consumption
satisfied according to consumer desires (as opposed to supply availability)
and the relative importance of consumption in any period versus consumption
in other periods and overall capital development, we can obtain quite different
results. But our range of choice is limited by initial stocks, productivities
IV-19
and the investment structure; these constraints cannot be circumvented.
Our reference, or baseline, solution ought to provide an idealized
illustration of the Soviet economy's potential at the beginning of 1929.
This would mean developing a solution which maintains total real consumption
at 1928 levels, preferably increasing steadily from that base. Further, it
would mean simultaneously building a capital stock by the end of our simulation
horizon in no way inferior to that actually achieved in Soviet experience,
and perhaps even superior to that standard. As should be clear from the
structure of our model, these two goals, higher consumption and higher
capital stock, must ultimately conflict. Hence developing the reference
solution is a process of varying the accessible parameters with a fine sense
of this balance in mind.
Developing the Reference Solution
We present a fairly detailed explaination of how the reference solution
was developed because the process is typical of how all the solutions
discussed below were developed. The cycle of solving the model, examining
the results, and adjusting the parameters for another go, is tedious, but
critical to the successful use of a model of this type. The user learns
how the model reacts, and develops confidence through the reasonableness of
those reactions (or goes back to respecifying the model if those reactions
are inappropriate). Hence we describe the procedure once, leaving the reader
to infer that the same procedure was used to develop solutions for the
historical scenarios on Chapter 6 below.
The logical starting point in a search for a reference solution is to
set the weights on all terms in the objective function equal to 1.0 in the
IV-20
final period. This implies raising the weights on consumption in earlier
periods by the appropriate power of the social discount rate, 10%, so that
consumption in 1929-1930 has a weight of 2.59. Further, the average propen-
sity to consume parameters were set at .5 for the first four periods, and
.7 for the last two periods. Thus real wages were only permitted to be half
nominal wages in the first period, etc. The fraction of consumption satisfied
in fixed proportions was .1, .1, .7, .9, .95 and .95 for periods one through
six respectively. Our previous experience with the model allowed us to
recognize that the solution would permit higher real wages, with consumption
more in line with peasant and worker preferences, in later rather than in
earlier periods. Table IV-9 lists these control parameter values for this
solution, #1, and for the other solutions leading up to and including the
reference solution.
The overall performance of the solution is quite good (see Table IV-10).
Over the 12 years, capital stock increase by a factor of 5, output by a
factor of 31/2 consumption by 21/2. However, the detailed results are clearly
unsatisfactory. There is a great deal of unemployed labor in the first four
periods, rising to almost 20% in the fourth period. More importantly, new
fixed capital is brought into production in sectors where the model finds it
least costly to invest, rather than in those sectors to which the Party
attached highest priority. Notice that the capital stock for the Government
Services sector is nearly half the total. If one examines Appendix C Table 11
one sees that investment in this sector requires a smaller proportion of
output from the Producer Goods sector than any other, and, as Producer Goods
Table IV-9. Parameter Values for Simulations Leading to the Reference Solution
Objective Function WeightsCapital Stocks, Jan. 1, 1941
Agriculture, Field CropsAgriculture, Livestock ProductsElectric PowerProducer Goods IndustryConsumer Goods Industry
Transport and CommunicationsConstructionHousingTrade and DistributionGovernment Services
Consumption1929 and 19301931 and 19321933 and 19341935 and 19361937 and 19381939 and 1940
Consumption ParametersAverage Propensity to Consume *
1929 and 19301931 and 19321933 and 19341935 and 19361937 and 19381939 and 1940
Fraction in Fixed Proportions *1929 and 19301931 and 19321933 and 19341935 and 19361937 and 19381939 and 1940
* These parameters were set identically for workers and peasants.
IV-23
is the limiting sector for investment, Government Services capital is the
"least expensive" capital that the model can produce. As the weighting of
the objective function values each sector's capital stock equally, the
optimizing algorithm's choice is algebraically impeccable.
In Solution #2 we chose to deal with the problem of the sectoral
distribution of investment. This was done by making a subjective estimate
reflecting the relative priority assigned to each sector by Soviet authorities
in the 1930's. The ten sectors can be grouped quite easily: first priority
went to the heavy industrial sectors, Producer Goods and Electric Power;
second rank went to the two agricultural sectors, field crops and livestock,
along with Consumer Goods (light industry), Transport and Communication, and
Construction; while Housing, Trade and Distribution, and Government Services
were put in a third category. We lowered the weights on the third group to
zero, that is, removed them from the objective function entirely. This
means that only enought capital stock will be built for these sectors to
enable them to provide the output required by exogenous demands and the needs
of more important sectors; in a sense these sectors will be dragged along
by the momentum in the model solution. The weights for the first group
were set to 3.0, making investment here more important than consumption.
The weights for the second group were set just above 2.0, making investment
in them more important than consumption in all but the first two periods
(see Table IV-9).
The summary report for Solution #2 is given in Table IV-11. The same
gross levels of investment, capital stock and output are achieved as in
IV-24
Solution #1 (Table IV-10). However, the sectoral distribution of investment
has shifted sharply away from Government Services to Producer Goods, whose
capital stock now comprises nearly half the total at the beginning of the
post-terminal period.
The shift has not been achieved without cost. The new intended pattern
of investment is more out of line with the initial structure of the economy
than that in the previous solution. This result is not unexpected, as a
major structural shift was a conscious policy goal of all Soviet planners.
In our model the process of structural change shows up most strongly in the
solution values for periods three and four (1933-1936), as a review of the
implementation of investment in the model will show. KAPROST requires
two periods (four years) of investment to produce new capital stock. There-
fore, as stressed earlier, capital additions in periods one and two had
already been exogenously fixed. Only in the third period (based on invest-
ment plans chosen by the solution algorithm in the first period) are additions
to capital stock responsive to the objective function weights. In theory
it might take the third period alone, or the third period and any number
thereafter, for structural change to be completed, that is, for capital
stock proportions to be shifted into a pattern compatible with the long-run
growth path implicit in the objective function weights and constrained by
the input-output and investment relationships in the model. In practice,
it would appear to require two periods, the third and the fourth. Solution #2
and all subsequent solutions in this chapter and in chapter six, show the
highest unemployment and lowest consumption in periods three and four,
with marked improvements in periods five and six. In particular, in Solution
IV-27
#2, Table IV-11, unemployment in these periods rises above 20%, and peasant
consumption falls sharply from Solution #1.
In order to improve consumption in the two middle periods, a more
detailed examination of the second solution is required. We found that
capital stock utilization in several sectors which contribute to consumption
in fixed proportions, but not to investment - Housing in particular -
was 100%. This bottleneck, would not be removed by raising the objective
function weights. Rather, it is best resolved by lowering the fraction of
consumption which must be satisfied in fixed proportions. This "frees up"
the limited housing stock, which may now be distributed among the larger
number of workers which we expect the model to hire.
Solution #3, Table IV-12, shows the effect of lowering to 0.5 the
fixed proportion component of consumption in periods three and four. In
addition the weight on capital stock in the Electric Power sector was raised
to 3.5 to see if investment could be switched away from the Producer Goods
sector (all control parameter values are listed in Table IV-9) . This second
objective was clearly accomplished, as the largest sectoral capital stock is
now Electric Power.
Examining the consumption and unemployment levels in solution //3 we
see that our selection of control parameters was correct, even if not the
best that could be hoped for. Unemployment is slightly higher in period
three as compared to Solution #2 (24.3% vs 23.7%) but appreciably lower in
period four (22.0% vs 26.8%). Workers consumption for 1933-34 and 1935-36
changes similarly - 15 billion rubles vs 15.4 billion rubles, and 24.8
billion rubles vs 21.7 billion rubles. However, the peasants are virtually
IV-31
unaffected, and detailed examination reveals that capital in both agricul-
tural sectors is fully utilized. Peasant employment is limited by a lack of
farm implements. We decided to attack this problem first by raising the
weight on capital stock for the two agricultural sectors and then by raising
the weight on peasant consumption in the third and fourth periods.
Solutions #4, #5, and #6 are reproduced in summary Table IV-13,
IV-14, IV-15. The weights on capital in both agricultural sectors move
up to 2.5, 2.6 and 3.0, respectively, while that on capital in Electric
Power is lowered to 3.3, 3.2 and 3.2. In Solution #4, this change brings
about a fall in unemployment and a rise in peasant consumption in periods
three and four, when compared to Solution #3. However, Solutions #5 and #6
show no further changes in either employment or consumption. Similarly, the
initial fall in the weight on capital stock in Electric Power causes a drop
in the final capital stock in that sector and a rise in Producer Goods
capital. Solutions #4, #5 and #6 are virtually identical, indicating that
we have found a point on the edge of our convex, feasible set which is
insensitive to those parameters over the ranges we have tested.
As we did not wish to raise the priority of agriculture above that of
the heavy industry sector, we proceeded to the second strategy: increasing
the weights on third and fourth period peasant consumption directly. In
Solutions #7, #8 and #9 we raise these weights to 2.5 and 2.0, 3.0 and 3.0,
and 3.2 and 3.0, respectively. The summary reports for these solutions are
presented in Tables IV-16, IV-17 and IV-18.
Compared to Solution #6, Solution #7 shows little change. However, by
IV-35
raising the third and fourth period consumption weights to 3.0 in Solution #8
we cross a threshhold, and succeed in eliminating unemployment in period
four (1935-36). This results in a sharp rise in fourth period peasant
consumption (13.3 to 24.7 billion rubles), at the cost of a small drop in
final capital stock (339.2 to 330.9 billion rubles). Finally, in Solution #9,
by pushing the weight on third period peasant consumption to 3.2, we eliminate
unemployment in that period also, and raise peasant consumption in 1933-34
to reasonable levels. Note here the cost is a further drop in final capital
stock of 4 billion rubles, and a loss of some fourth period consumption.
Solution #9 has most of the properties we listed above as desirable in
a reference solution. It is clear, however, that Producer Goods industry
as a whole would in practice require a larger stock of fixed capital than
Electric Power alone, even given Lenin's stress on electrification.
Our final adjustment, therefore, involves shifting the weights for capital
in these two sectors to 3.1 and 3.0 respectively. The result is a balanced
reference solution which deserves detailed examination.
A Balanced Reference Solution
We have now uncovered an expansion path, consistent with all the model's
parameters, that exhibits a realistic balance between consumption and
accumulation while at the same time drawing into use most of the economy's
productive capacity. Its details are set forth in the accompanying printout,
in the form of five post-solution reports. The five-page summary report
lists objective function elements and displays a number of totals over all
ten sectors for each of the two-year time periods: 1929 and 1930, ....,
1939 and 1940.
IV-36
It then shows gross output levels produced with old-technology capital by
each sector in each period, and similarly the output produced with new-
technology capital by each sector in each period. Another pair of panels
shows the expansion, sector by sector and period by period, of new-technology
stocks of fixed capital. These are primal solution values.
The summary report continues with dual values for capital stocks, by
technology, sector, and period, and for sectoral gross output levels, by
period. These values can be interpreted as indicators of the extent to
which a unit increase in the constraint could increase the value of the
objective function in that period. Large dual valuations are a sign of
bottlenecks. Finally, the summary report tabulates the deliveries to
optional consumption, by sector and period, as a measure of the extent to
which strain in the economy requires deviations from the standard pattern
of consumption by peasants and workers.
For each period the model generates a full set of flow-table entries
like those presented for 1928 and 1933 in Chapter III. These are displayed
for the first period as an interindustry flow report (for the northwest
quadrant), a final demand report (for the northeast quadrant), and a value
added report (for the southwest quadrant). Finally, there is a resource
utilization report giving details on capital and labor utilization in each
sector in each period. For periods after the first, the interindustry and
value added tables are not included in the accompanying material.
IV-56
The central message of this reference solution is that the Soviet economy
in the 1929-1940 period contained a very substantial potential for output ex-
pansion, under problem-free conditions, without much need for belt-tightening.
Given the base period situation, no model is needed to suggest such a conclusion,
since the resources and growth possibilities were obviously at hand. What the
model does make possible, however, to the extent that i ts assumptions and
calibrations are accepted, are at least some crude quantitative estimates of
the degree and kind of expansion that might have been possible in the absence
of any of the negative developments Chat actually occurred.
In summary form, the expansion path is marked out in the data of Table IV-19
and Chart IV-1. The economy's gross output in period six (1939 plus 1940) could
have been some 3.8 times as large as in Che base period (twice 1928), Deducting
interindustry deliveries, the GNP could have risen from an index of 100 to an
index of 424. Within these aggregates, consumption would have increased some
2.7-fold while investment was rising from an index of 100 to an index of 824.
The stock of fixed capital at the beginning of 1941 would have been five times
larger than i t was at the beginning of 1929. Deliveries to final demand from
the two agricultural sectors would have risen quickly, according to this
reference solution, from a two-year total of 15.0 billion rubles in the base
Table IV-19. Reference Solution Values for Gross Output, GNP, and Components, By Period,
absolute data in billions of rubles at base-period prices6th
Twice 1st 2nd 3rd 4th 5th 6th Period1928 Period Period Period Period Period Period Index(a)
Total gross output 106.0GNP:
Agricultural output 15.0Industry and construction 16.9Services 19.7Total GNP 51.6
Consumption 30.6Investment 11.8Government services 5.9Exports 1.2
End-of-period stock of fixed capital 65.7Percent Increases per period:
Total gross outputAgricultural outputIndustry and constructionServicesTotal GNPConsumptionInvestment
Percent shares in GNP:Consumption 59.3Investment 22.9
Absolute Increments per period:Fixed capitalGNP
Incremental capital/output ratio
122.2 149.3 182.1 258.1 331.8 405.8
19.526.419.365.2
34.420.07.71.774.6
15.330.056.2-2.026.412.469.5
52.830.7
21.439.620.181.1
33.233.710.12.3
102.5
22.29.750.04.124.4-3.568.5
40.941.6
8.9 27.913.6 15.90.65 1.75
30.351.220.4101.9
41.041.812.53.0
122.1
22.041.629.31.525.623.524.0
40.241.0
19.620.80.94
26.889.129.7145.6
40.679.514.93.6
192.5
41.7-11.674.045.642.9-1.090.2
27.954.6
70.443.71.61
27.3107.444.0178.7
66.779.417.34.2
248.1
28.61.920482264-0.1
37.344.4
29.7135.753.4218.8
82.497.219.74.8
327.4
22.38.826.421.422.423.522.4
37.744.4
383
198803271424
269824334400498
55.6 79.333.1 40.11.68 1.98
(a) With the base period (twice 1928) taken as 100.
IV-58
CHART IV-1. REFERENCE SOLUTION VALUES FOR GROSS OUTPUTAND RELATED SERIES, BY PERIOD, IN BILLIONSOF RUBLES AT BASE-PERIOD PRICES
Twice 1923 1929+1930 1931+1932 1933+1934 1935+1936 1937+1933 1939+1940
IV-59
period to a total of 30.3 billion in period three, then fallen off slightly
to 27 billion in the next two periods and 30 billion in the sixth period —
double the initial level. Clearly so rapid a degree of improvement in agri-
culture is empirically rather unlikely. Over this same period the output of
the service sectors would have been rising from an index of 100 to an index
of 271, with the growth concentrated in the last three periods. Deliveries
to final demand by the two industry sectors and the construction sector would
have grown rapidly and continuously, reaching by the sixth period a level
eight times higher than the base-period level. Thus would the optimistic
hopes of the system's directors have been realised.
Expansion on this scale would have required a drastic shift away from
consumption and toward investment. Though the absolute level of consumption
would not have fallen, and in fact it would have risen from 30 to over 80
billion rubles, the share of consumption in GNP would have declined from .59%
in the base period to a low of 282 in the fourth period and an apparently
stable share of 37-382 in the last two periods. Conversely, the absolute
amount of investment would have risen from 12 billion rubles to almost 100,
and the investment share in GNP would have increased from 232 to 552 in
period four and 442 thereafter. This 3tress on capital formation would of
course have raised capital stocks very substantially, to more than three
times their size at the beginning of 1929. Comparison of period-to-period
increments in fixed capital and in GNP discloses erratic changes in the incre-
mental capital/output ratio, which rises from 0.65 in the first period to
1.98 in the sixth. These are very low values, by comparison with recent LDC
experience, especially in view of the sectors being stressed. Again it is
evident that this reference solution embodies optimistic parameters.
Given our way of projecting out to 1940 the Gosplan intentions for
IV-6Q
final-demand deliveries to exports and to government services during 1929-1933,
their shares of GNP show modest changes. Exports made up 2.4% of Soviet GNP
in 1928, and this share rises to 2.9% in periods two and three, only to fall
off to 2,2% in the sixth period. Similarly, government services account for
11.4% of the 1928 GNP, a share that would have risen to 12.5% in period two
and then declined to 9% by the sixth period.
Inspection of the dual valuations in this reference solution reveals a
crucially binding constraint in the construction sector in the second period.
It was here that limitations on the amount of fixed capital construction
begun before the plan period but not yet completed, together with the limita-
tion imposed by a two-period gestation period for new fixed capital, would
have placed an insurmountable barrier to further advance. Once past this bottle-
neck in the second period, the model generates construction-sector output
without strain and the dual values fall to negligible levels. There is also
an indication that electric power in the second period was under considerable
strain, but that the strain would have been relieved thereafter. The pattern
of dual valuations varies noticeably under experimentation since these values
are quite sensitive to the linear structure of the model. They are not to
be relied on for a precisely detailed account of sectoral and temporal strains
in this hypothetical reconstruction of a problem-free expansion path.
Through further juggling of weights on individual terms in the model's
objective function (peasant consumption in each period, worker consumption in
each period, and sectoral capital stocks in the final period), we could have
smoothed out some of the sharp fluctuations evident in this reference solution,
and tailored the capital pattern more closely to assumed Party intentions.
The analytic gain would be negligible. Sharp shifts are inherent in linear
programming. As long as one can assume convexity in the model's structure,
IV- 61
a linear combination of two neighboring solutions will be feasible by definition,
so gradual transitions are implicitly available, even if not strictly optimal.
This particular reference solution, for example, builds 161 billion rubles
worth of new-technology capital in the producer goods sector by January 1,
1941, and only 10.1 billion billion rubles worth for the electric power
sector. Examination of Table IV-18 above, recording Solution #9, shows
however that by reversing the weights on capital stock for these two sectors
we can reverse the proportions, creating 114 billion rubles worth of capital
stock in electric power, and 53.4 billion in producer goods. In practice,
the combined total of about 170 billion rubles could have been divided
between these two sectors in almost any desired proportions, within the
overall set of constraints forming the model. The linear programming model
tells us about the zone which bounds feasible territory; it need not be
interpreted as marking out a single, unique, fine-scale "optimal" solution.
It should also be noted that the feasibility of this problem-free refer-
ence solution is affected by its high degree of aggregation. A linear program-
ming approach treats the activities within each sector as completely homogeneous
so that intrasectoral flows are not constrained by technical coefficients, and
intermediate goods are completely substitutable for each other. "It is not
known in advance whether aggregation of sectors, on balance, overstates or
understates the potential achievements of a system. Aggregation permits full
substitution within sectors and thus may overstate production potential. On
the other hand, if as the result of aggregation the average input requirements
imposed in a particularly fast-growing part of the economy are higher than if
that sector were treated separately, the effect is to penalize growth." More-
IV-62
over our use of two-year time periods "...implies complete substitution within
each period. This provides "an additional degree of freedom in breaking
bottlenecks, which is not in fact present..." (though changing from one year to
two years cannot have a decisive influence on evolution over a twelve-year
horizon). "On the other hand there may be compensating disadvantages in tine
aggregation since it forces a kind of synchronization among sectors in each
aggregated time period which in actuality need not be present." Quotes from
R. S. Eckaus and K. S. Parikh, Planning for Growth (1968), p. 154.
In this chapter we have sought to obtain a multisectoral, multiperiod
picture of the way the Soviet economy might have expanded from 1928 through
1940 if the economy's potential had been drawn on as the Party intended and
no difficulties had intervened. Our reference solution appears to mark out
the upper limits of the possible. In historical actuality, however, the
economy suffered a number of serious blows, each leaving its mark on parts
of the system. After sketching the key developments in the next chapter,
we use the model in Chapter VI to decompose the actual record in order to
measure the impact of these events, thus accounting for the differences
between our reference solution and the actual course of events.
VI-1
VI. DECOMPOSING THE RECORD TO MEASURE THE IMPACT OF MAJOR EVENTS
Having compiled broad evidence concerning several major economic
developments in the USSR during the 1930s, we are now ready to insert these
data into the quantitative framework provided by our reference solution for
the 1928-1940 period. Each of the events had an obvious bearing on specific
aspects of the economy. Taken all together, the developments ought to account
for the observed differences between the actual path of the Soviet economy and
the problem-free path spelled out in our reference solution. In theory it
should be possible to derive quantitative estimates of the relative importance
of the individual events in affecting overall results. Ideally, therefore,
we might hope to decompose the actual record, identifying and measuring the
major causal factors at work. This would not be counterfactual history, trying
to estimate what did not happen, and thus should not offend those who, like
Edward Hallet Carr, call for a focus on the interacting complexities of events
as they in fact occurred.
For several reasons, caution is in order concerning the precision of
results to be expected. Soviet statistics for the 1930s were collected under
conditions that tended to produce exaggerations, and were reported in ways that
distorted the record. Even if these defects in the evidence can be overcome,
valuation issues during a period of marked structural transformation introduce
ambiguities that make single-valued answers impossible. In addition, our ten-
sector, six-period model is quite highly aggregated, and though its rigidities
are reduced by a number of technical features, results generated by the model
reflect its oversimplified design. Models with other properties would tell a
different story using the same evidence. We have found, nevertheless, that
extensive experimentation discloses some central tendencies that stand up under
VI-2
a variety of alternative assumptions and over a wide range of parametric
variations. The numerical pattern that emerges is, fortunately, fully in
accord with the qualitative impressions that prevail among most students of
the period. We assert, therefore, that our results provide useful illustrative
detail that throws illuminating light on the forces at work in a very important
and interesting period of structural change.
Of the seven developments in the Soviet economy described in Chapter V
above, the investment drive has already been built into our reference solution,
along with three of the four years of transport strain. The influence of two
other developments: (a) inflation, rationing, and consumer goods shortages,
and (b) political purges and mass terror, was indirect and not easily measured.
We focus, therefore, on the effects of the world depression, of rearmament, and
of agricultural collectivization, each of which can be easily translated into
linear constraints suitable for inclusion in KAPROST. The first two were
clearly exogenous influences; the Soviet government had no control over
the world-wide depression, and little choice but to rearm. However the
decision to collectivize, and its attendant effects, were solely the result of
internal policy decisions. Hence we will examine the impact of each of the
exogenous influences separately, and then combine them to evaluate the total
impact of external affairs on the Soviet economy. We then look at collecti-
vization by itself, as most scholars would agree this was the most critical of
Stalin's policy decisions. Finally we combine all three impacts, hoping to
see a model solution which tracks historical reality fairly well. The difference
between this composite solution and the actual record we attribute to those
influences we have not been able to quantify, and the limitations of our basic
model data.
As noted in Chapter IV, even with the constraints imposed by the model,
an essentially infinite variety of solutions exist. This implies that the results
we present are unavoidably influenced by our judgment as to what would be
VI-3
appropriate and reasonable. We have tried to guide our choices by those
goals which were nearly universal among Soviet policy makers: economic
development through capital investment, with consumption as a secondary
objective.
The Trade Impact of World-Wide Depression
Soviet foreign trade statistics for the 1930s are even more compli-
cated than those of other countries in normal times, as the economy was
undergoing structural change, and price movements were very marked during the
world depression. Many alternative weight years deserve consideration and
index number issues abound. Straightforward, unambiguous computations are not
possible. We take advantage of Michael R. Dohan's painstaking research, and
use his aggregate indices as presented in Table V-4 above.
Table VI-1 repeats Dohan's estimated volume index for Soviet foreign
trade, along with the officially compiled series. Both indexes are derived from
series in foreign-trade rubles, since the USSR does not release data on exports
and imports in domestic prices. The official series corrects for exchange-rate
changes but otherwise reflects current prices. Professor Dohan's series meticu-
lously applies price weights of 1927/28 to quantity data with, fuller and more
consistent coverage than previously employed. As discussed more fully in his
basic report, the corrected series show a smaller drop in imports and exports
after 1931 than displayed by the official series, since the latter reflect both
reduced quantities and lower prices.
While it is literally true that the money value of commodity trade
shrank in the 1930s for both price and volume reasons, the whole KAPROST project
is cast in constant-price terms. Hence we use Dohan's volume indexes to reflect
changes in physical imports and exports over the 1928-1940 period, measured at
VI-4
TABLE VI-1. Volume indexes for Soviet exports and imports,
by year, 1928-1940, 1929 = 100
YEAR
E X P O R T S
OFFICIAL DOHAN
I M P O R T S
OFFICIAL DOHAN
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
85.7
100.0
112.2
87.8
62.3
53.7
45.3
39.8
33.6
40.8
31.7
14.4
33.1
77.5
100.0
149.5
164.0
128.2
124.6
117.9
105.6
83.7
87.3
78.0
(40.2)*
(138.3)*
107.4
100.0
120.2
125.5
80.0
39.6
26.4
27.4
35.1
33.1
35.5
24.3
35.5
101.1
100.0
123.1
146.1
107.0
66.2
59.7
66.8
68.0
62.2
69.2
(46.8)*
(68.4)*
Sources: Official index derived from absolute figures at the 1950 ruble
exchange rate in MVT, Vneshniaia Torgovlia SSSR ZA 1918-1940 gg
(1960), p.17. Dohan indexes from Michael Dohan and Edward Hewett,
Two Studies in Soviet Terms of Trade, 1918-1970 (Bloomington, Indiana:
Indiana Univ. Int. Development and Research Center, 1973), pp 24 and 27
* Our estimate, as described in text.
VI-5
constant prices. Our reference solution incorporates hypothetical export and
imports obtained by extending planners' intentions in a problem-free environ-
ment; we want to replace these trade figures with quantity flows reflecting
the physical changes that occurred. The additional impact of worsened terms
of trade is analyzed by Michael Dohan; no attempt to enter this range of issues
is made here.
The Dohan series ends in 1938, while our model requires data through
1940. The extension was accomplished by applying the proportionate changes in
the official series for 1939 and 1940 to the last number in Dohan's series.
These estimates are in parentheses in Table VI-1.
While in time it would be possible to derive disaggregated trade series
compatible with KAPROST's four trading sectors, we have initially limited our-
selves to applying aggregate indexes equally to trade in each sector. The model
requires exogenous estimates of exports by sector. Table IV-2 presents these on
a yearly basis for the four exporting sectors, obtained by applying the Dohan
index in Table IV-1 to the 1928 trade numbers in Table III-5. These are summed
in Table VI-3 to yield two-year exports compatible with KAPROST's structure.
Imports in KAPROST are derived first as a ratio of achieved output, and
second, if a net export surplus still exists, as discretionary increments to
critical sectors. We would like to limit total imports to those amounts given
in Tables VI-2 and VI-3. As the model requires net exports plus foreign trade
credits to balance in all periods, by setting foreign trade credits equal to net
imports, this result is achieved. Table VI-3 presents all the data needed for
the trade scenario.
Characteristics of the Trade Solution.
Application of the trade data with no further modifications has a
crippling effect on the KAPROST solution. Capital stock falls to 25% of the
reference solution, sixth period output to 30% of the reference. Examination
VI- 6
TABLE VI-2. Exports by Year and Sector, Total Imports
AGRICULTURE,FIELD CROPS
YEAR
1928 76
1929 98
1930 147
1931 161
1932 126
1933 122
1934 116
1935 104
1936 82
1937 86
1938 76
1939 31
1940 105
(Millions of Rubles
S E C T O R
E X P O R T S
AGRICULTURE,LIVESTOCKPRODUCTS
230
297
444
487
380
370
350
313
248
259
231
92
PRODUCERGOODSINDUSTRY
263
339
507
557
435
423
400
358
284
296
265
106
318 364
)
CONSUMERGOODS
INDUSTRY
50
65
96
106
83
80
76
68
54
56
50
20
69
EXPORT
TOTAL
619
799
1194
1310
1024 .
995
941
844
669
697
623
249
856
IMPORT
TOTAL
792
783
965
1144
838
519
468
524
533
487
542
371
542
1928 Values are from reconstructed input-output Table III-5.
Other values obtained by applying the Dohan index, Table V-l to the 1928 values
VI-7
TABLE VI-3. Historical Trade Data for KAPROST.
(Millions of rubles)
S E C T O R
E X P O R T S
PERIOD
1929-30
1931-32
1933-34
1935-36
1937-38
1939-40
AGRICULTURE,FIELD CROPS
245
287
238
186
162
136
AGRICULTURE,LIVESTOCKPRODUCTS
741
867
720
561
490
410
PRODUCERGOODSINDUSTRY
846
992
823
642
561
470
CONSUMERGOODS
INDUSTRY
161
189
156
122
106
89
EXPORT
TOTAL
1993
2334
1936
1513
1320
1105
IMPORT
TOTAL
1748
1982
987
1057
1029
913
FOREIGNTRADE
CREDITS
-245
-352
-949
-456
-291
-192
Two year total from Table VI-2, except for:
Foreign Trade Credits = Import Total - Export Total.
VI-8
of the detailed output showed clearly that, given the required import ratios,
the historical level of imports will not support any greater level of economic
activity. As this output is significantly below that which actually occurred
during the 1930s, an adjustment is obviously necessary.
The USSR made some emergency imports of raw cotton, steel, copper, and
a few consumer goods at the start of this period, and many of the early, one-
time imports of machinery and equipment embodied advanced technology that played
a key role in subsequent industrial expansion. But while world depression
clearly reduced planners' flexibility, there is no evidence that any significant
loss of output resulted from an inability to import goods after 1932. Some level
of required imports should be retained to assure that imports are distributed
among all importing sectors, but not so much as to significantly bind the system.
To make this descriptive result precise, the required import ratios were lowered
proportionately until: the largest dual valuation (shadow price) on the balance
of payments constraint is of the same magnitude as the dual value of other con-
straints; and that each period shows some amount of optional imports, but in some
period that number is relatively small.
Table VI-4 presents a comparison of the effect of imposing varying
fractions of the original required import ratios (Appendix C, Table 14) on the
reference solution. The trade data from Table VI-3 are used, but all other data
and parameters are as in the reference solution described in Chapter IV. Note
that as none of the objective function weights have been changed, the objective
function value is comparable across solutions.
Examination of Table VI-4 indicates that we want to impose between 20
and 25% of the original required import ratio. Between these two values, the
dual value of the most stringent balance of payments constraint rises sharply,
VI-9
Table VI-4. Effects of Varying the Required Import Ratios(Billions of Rubles)
Percent ofRequiredImport Ratio
100%
75%
50%
25%
20%
10%
0%
ReferenceSolution
(1)TotalCapitalStock
77.8
118.0
186.8
279.7
290.8
292.7
294.7
294.0
(2)6th PeriodTotalOutput
129.3
169.9
246.9
377.0
407.0
408.4
409.9
405.8
(3)LargestDualValue
219
211
210
?96
10
10
10
9
(4)LeastOptionalImports
0.0
0.0
0.0
0.0
.085
.497
.913
.236
(5)Value ofObjectiveFunction
807.2
934.5
1040.9
1133.2
1207.4
1211.6
1215.8
1219.4
(1) New technology capital stock, January 1, 1941.
(2) Total 1939 + 1940 output.
(3) Highest dual valuation on balance of payments constraintover all six periods, Objective-function increase per unit rise inconstraint.
(4) Smallest total of optional imports over all six periods.
(5) Calculated value of the objective function, summing consumption oversix periods and Jan. 1, 1941 capital stocks over ten sectors.
VI-10
and we begin to see a sharp deterioration in capital stock and output. Note
that with no required imports, or only 10% of the original ratio in force, the
trade scenario is comparable to the reference solution. Hence it is only by
diminishing the flexibility allowed to the optimizing algorithm (or Soviet
planners) that world depression has an important impact on the USSR.
The summary report for the "20% solution" is reproduced in Table VI-5,
and for the "25% solution" in Table VI-6.
The Effects of Rearmament.
Rearmament was felt in the Soviet economy only during the last half of
the 1930s. In evident recognition of the fact that military hardware cannot be
produced without a sizable heavy industry sector, the initial investment surge
focused on producer goods industry. However, military hardware requires military
manpower, and, then as now, the exact division of military expenditures among
branches of industry is difficult to ascertain. Hence we are forced to estimate
some components of military demand from manpower levels, assigning the remainder
to hardware.
Table V-6 above presents the Defense expenditures and Armed Forces man-
power levels for the Soviet Union over our simulation horizon. Note that for the
years 1928-1932 both figures are nearly constant (or unavailable). As a result
we decided to take this minimum force level, 560,000 men and their associated
costs, as implicit in the economy. Thus we assume no changes due to defense in
1929-1932, the first two simulation periods, and we subtract expenditure and
manpower requirements for these men from all later years before calculating the
impact of rearmament.
Where, then, is defense spending to be applied to the economy? We
assumed it would occur in basically two ways. First, troops need food, clothing,
shelter and other basic equipment, both for themselves and their dependents,
VI-13
which implies consumption demand and some capital investment not very different
from that of other sectors of the economy. Second, especially in the late
1930s, an army required deliveries of trucks, tanks, artillery, planes, ships,
etc., all of which come directly from the Producer Goods sector.
In order to provide for the soldiers, we simply employed them through
the Government Services sector of the economy. As shown in Table VI-7, we
take average armed forces manpower in model periods 3 through 6, subtract 560
thousand to determine the increase above minimum force levels and then, using
the output-labor ratio for new technology in the Government Services sector,
determine the increase in demand for such output needed to employ those men.
We add this to historical levels of Government Services output to obtain the
new value of government demand for Government Services in KAPROST. Note this
will not only employ labor equal to the growth in the armed forces, but will
also require the model to build more capital stock for that sector (government
office buildings and army barracks are presumed more or less equal). The wages
paid to the workers will generate increased demand for consumption (soldiers and
workers presumed more or less equal). Note that while army and civilian wages
were certainly not equal, it is real demand for goods we are trying to generate
by this scheme.
The demand for military hardware will be represented by a government
demand for deliveries from the Producer Goods sector. The magnitude will be
the total defense spending less that amount accounted for through increased
government demand for output of Government Services and consumption demand by
the soldiers so employed, as described above. These calculations are shown in
Table VI-8. Note the increased consumption demand by soldiers is obtained by
adjusting the wage payments by the average consumption coefficient, .5 in
periods 3 and 4, .7 in periods 5 and 6.
VI-14
TABLE VI-7. Increased Demand for Government Services
Due to Increased Size of Armed Forces
PERIOD
3
4
5
6
(1)
ARMED FORCES
SIZE
913
1184
1473
3534
(2)
ARMED FORCES
INCREMENT
353
524
913
2974
GOVERNMENT SERVICES(3)
INCREMENT
2871
5075
7425
24188
(4)HISTORIC
9867
11119
12664
14823
OUTPUT(5)
TOTAL
12738
16194
20089
39011
(1) Two year averages from Table V-6, in thousands
(2) = (1) - 560. 560 is the average 1929-1932 force level, in thousands.
(3) = (2) x 8.13315, thousands of rubles. 8.13315 is the output-labor
ratio for Government Services new technology production.
(4) Two year total from Table V-9, the row for Government Services sector,
in thousands of rubles.
(5) = (3) + (4), in thousands of rubles.
VI-15
One further adjustment is required. For the first two periods we
reduce government demand for Government Services to historical levels. In
keeping with the capital stock and capital productivity adjustments described
in Chapter IV we adjust both old and new technology output-capital ratios
in this sector to achieve exactly the historical output in periods 1 and 2
(Table VI-9). In the reference solution we had used linearly interpolated
values from 1928 actual and 1933 planned levels.
Developing a Solution with Rearmament.
As with the trade scenario, direct application of the rearmament data
to the reference solution does not have a salubrious effect; in fact, an in-
feasible solution results. All of the competing demands on output cannot be
satisfied simultaneously, so we proceeded in good Soviet fashion to lower
consumer demand. By lowering real wages to 50% of nominal wages, and reducing
the fraction of consumer demand satisfied in fixed proportions for the last
two periods to half, the model solved. This type of improvement is what one
would expect in a real economy: guns are possible when the amount and variety
of butter is reduced.
The solution still showed some undesirable characteristics: con-
sumption was too low in some periods, and in others included too much of one
sector's output but not enough of others; capital stock was low, and could have
been distributed better; etc. Further adjustments were made to the real wage
and consumption proportion parameters, and to the objective function weights for
capital in agriculture (livestock) and producer goods, as well as third period
peasant consumption. Table VI-10 lists the differences in parameter settings for
this solution versus the reference. Table VI-11 presents our "rearmament only"
solution.
VI-16
(1)
TABLE VI-8
Demand for Military Hardware
(2) (3)Military
(4)
Period
3
4
5
6
Total DefenseSpending
6440
23069
40632
95933
GovernmentServices
2871
5075
7425
24188
ConsumptionDemand
539
991
2180
6807
MilitaryHardware
3030
16995
31027
64983
All values in thousands of rubles.
(1) Two-year totals from Table V-6
(2) From Table VI-7, Column 3.
(3) Table VI-7, Column 1, multiplied by the Government Services wage rate for
each period (Appendix C, Table 10, minus 907.3 thousand rubles (the basic
wage cost for 560 thousand soldiers in 1929) all multiplied by the average
propensity to consume, .5 in periods 3 and 4, .7 in periods 5 and 6.
(4) = (1) - (2) - (3), the government demand for deliveries from the Producers
Goods sector.
VI-17
Table VI-9
Adjustments to Government ServicesOutput-Capital Ratios, Periods 1 and 2
(1) GS Output 1929-1930 6786
(2) New Technology 222
(3) Old Technology 6564
(4) GS Old Technology Capital 1929-1930 5519
(5) Adjusted Old Technology Output-Capital 1.18935Ratio for Periods 1 and 2
(6) GS Output 1930-1931 9279
(7) Old Technology 6196
(8) New Technology 3083
(9) GS Available New Technology Capital 1929-1930 1079
(10) Adjusted New Technology Output-Capital 2.85728Ratio for 1930-1931
All values in thousands of rubles except lines (5) and (10).
(1) Two-year total, Table V-9
(2) First period new technology capital stock (246) from Appendix C, Table 6,
multiplied by first period new technology output-capital ratio (0.90313)
(3) = (1) - (2)
(4) From Appendix C, Table 6
(5) = (3) ÷ (4)
(6) Two-year total, Table V-9
(7) = (3) depreciated one period
(8) - (6) - (7)
(9) From Table IV - 5
(10) = (8) ÷ (9).
VI-18
Table VI-10
Control Parameter Settings: Rearmament vs. Reference
Parameter Rearmament Reference
Objective Function WeightsCapital Stock, 1/1/41Agriculture: LivestockProducer Goods
Consumption1933-34
2.83.2
3.4
3.03.1
3.2
Average Propensity to Consume1929-301931-321933-341935-361937-381939-40
0.40.40.40.40.40.4
0.70.7
Fraction of Consumption inFixed Proportions
1933-341935-361937-381939-40
0.30.40.80.9
Reference solution parameter values from Table IV-8.
Parameters not listed are identical for both solutions
0.50.50.950.95
VI-20
Superficially, the impact of rearmament on the Soviet economy
seems severe. Comparing Table VI-11 to the reference solution in Chapter IV,
final capital falls from 327 to 276 billion rubles, and unemployment and con-
sumption worsen in the final four periods. Despite a fair amount of effort, we
were not able to avoid the unemployment and consumption impact, especially the
severe third period result: 1933-34 unemployment of 23.2% versus 0% in the
reference; 1933-34 total consumption of 27.3 billion rubles versus 41.0 billion
rubles (twice the 1928 level of consumption is 30.5 billion rubles). The
capital stock, however does not include 109 billion rubles of military hardware
produced for defense (see Table VI-12 for the calculation of this figure).
Further total output in the sixth period (1939-40) is approximately equal for
both solutions: 405 billion rubles in the reference, 419 billion rubles with
rearmament. KAPROST seems to indicate that while rearmament had a mild cost in
terms of real income, it did not significantly halt development over the 1930s.
Clearly, if war had not intervened, progress in the 1940s would have slowed by
the reduced availability of productive capital, but the economic problems of
the 1930s cannot be blamed on the preparations for war.
VI-21
Table VI-12
Stock of Military Hardware, Jan. 1, 1941
Period
1933-34
1935-36
1937-38
1939-40
Total Milit
(1)Military-Hardware
Expenditures
3030
16995
31027
64983
ary Hardware, Jan. 1, 1941:
(2)Depreciation
Factor to1/1/41
.72723
.80869
.89927
1.0
(3)MilitaryHardware1/1/41
2203
13744
27902
64983
108832
(1) From Table VI-8, Column 4. millions of rubles.
(2) Appropriate power of (1-Depreciation Rate) for old technology
capital in the Producer-Goods sector.
(3) - (1) x (2).
VI-22
The Combined Impact of Trade and Defense
Of the three events whose impact we can quantify and introduce into
KAPROST, the deterioration of trade and the decision to rearm were driven
by exogenous factors. Hence it is of some interest to determine how much
these developments caused the Soviet economy to deviate from a more desirable
path. Table VI-13 combines the preceding adjustments in order to do just this.
All data modifications to implement the deterioration of trade (with 20% of the
required import ratios in force), and all defense expenditures have been in-
cluded. The consumption parameters and objective function weights are set as
in the rearmament solution just discussed.
In examining this solution four things ought to be noted. First, both
consumption and unemployment are maintained at reasonable levels - not as
desirable as the reference solution, to be sure, but not conspicuously poor
either. Second, while capital stock on Jan. 1, 1941 is significantly lower
than that achieved in the reference, this fall is illusory. As noted above,
the total does not include the military hardware produced. Third, sixth period
(1939-40) total output, at 401.6 billion rubles, is insignificantly less than
that achieved in the reference solution. Fourth, combined with the demands of
the military, the balance of payments constraint becomes sharply binding in the
sixth period, with a dual valuation of 284 to 1.
Collectivization in Agriculture
The effect of collectivization on agricultural output and capital stock
is modelled very directly in KAPROST. Actual historical levels of both production
and capital in agriculture are imposed as constraints on the model. We turn
first to estimating these values, then to adapting them to the model, searching
for a representative solution, and discussing the results.
VI-24
We have already described briefly the way that collectivization of
agriculture led to smaller herds of livestock and reduced availability
of animal draft power and livestock products, while at the same time raising
the share of agricultural output that was taken for off-farm use. The Soviet
diet, both urban and rural, deteriorated substantially. Detailed analysis is
beyond the scope of this study, but a few computations will illustrate some of
the changes and also supply data so that KAPROST can reflect their impacts.
The shrinkage of livestock herds was a very unfavorable development
but it had the incidental effect of releasing for human consumption some of
the grain and other feed that had been supporting livestock. Without attemp-
ting to trace consumption of vegetables, oilseed cake, etc., one can make crude
estimates of the shift in grain consumption on the basis of readily available
data. Before collectivization about half the total grain crop was consumed
directly as food while roughly a quarter of it went into feed for animals. The
big consumers were horses, especially those of working age, and hogs; smaller
amounts went to cattle, sheep and goats, and poultry. Using summary official
data for herd numbers together with feeding rates per animal as carefully
evaluated by Naum Jasny, Table VI-14 indicates the magnitude of the changes in
grain use by working horses, hogs, and cows (omitting young horses, beef cattle,
sheep and goats, and poultry). The first three columns show the size of the
absolute number, in millions of head, by which the January 1 animal count fell
short of the number on hand at the beginning of 1929, year by year through 1940.
For example, the 22.4 million working horses available in 1929 were reduced by
January 1, 1935, to 10.4 million, a decline of 12.4 million. After 1934 the
horse population rose again, reducing the shortfall. There was restoration in
the number of cows, too, and the number of hogs rose above the base-period level.
VI-25
Applying feeding rates of 450, 245, and 60 kilograms per animal per year,
column 4 shows the amount of grain, in millions of tons, that was released by
these declines in livestock herds. In a period when direct human grain con-
sumption was in the neighborhood of 35 to 40 million tons per year, something
like a fifth of it was made available this way. As Jerzy Karcz pointed out,
"...the entire increase in grain marketings between 1928 and 1933 can be
ascribed to the decline in livestock herds over that period. (The Economics
of Communist Agriculture, Bloomington, Indiana; International Development
Institute, 1980, p. 455).
Since labor was not a constraining input in agriculture, we can focus
on changes in fixed capital as a key factor associated with the output changes
reported in chapter V. Fixed capital in agriculture takes many forms: land,
livestock, barns and houses, wagons, plows, and other equipment, etc., and in
this project it has been subdivided between fieldcrop agriculture and livestock-
products agriculture in fairly summary fashion. It is important to note that in
Soviet agriculture at the beginning of the 1928-1940 period, draft power for
conducting all agricultural activity, in both divisions of agriculture, was
overwhelmingly animal power — horses together with oxen in the South and a few
camels in Soviet Central Asia. We have therefore assigned the value of the 32.6
million horses on hand at the beginning of 1929 — making up about 49% of the
total value of all livestock — to fieldcrop agriculture, leaving the cattle and
cows, hogs, sheep and goats, poultry, and bees to be part of the fixed capital
in the livestock-products division of the agricultural sector. Both stocks de-
clined and rose again, as shown in Table VI-14. The loss of animal draft power
was a major blow to agricultural operations, forcing the Party to provide first
imported and then domestically-produced tractors to fill the gap. In addition
a substantial volume of mechanized equipment in the form of reapers, harvesters,
VI-26
Table VI-14
Reduced numbers of working horses, hogs, and cows, and amount of grainreleased, by year 1929-1940, in millions of animals and metric tons.
Working Grainhorses Hogs Cows Released
On hand, 1 Jan 29 22.8 19.4 29.2
DecrementTo Jan. 1
1930
31
32
1933
34
35
36
1937
38
39
40
1941 8.1 -8.1 1.4 1.7
Sources: Derived from herd data in TsSU, Selskoe Khoziaistvo SSSR (1960), p. 263,
and feeding rates from Naum Jasny, "Soviet Grain Crops and Their Distri-
bution," International Affairs, Oct. 1952, p. 458. The ratio of working to
total horses was .732, .722, and .716 during 1927-29 (KTs 29/30, pp. 530-31).
FYP, II, 1, pp. 332-33 puts it at .696 for 1927/28 and .692 for 1932/33
(intended). The TsSU series for all horses is here multiplied by .7. See
also Jasny, Socialized Agriculture of the USSR, pp. 748-56, esp. p 751.
1.1
3.9
7.6
10.7
12.0
12.4
12.0
11.7
11.5
10.8
10.4
5.2
7.7
8.5
9.5
7.9
2.3
-6.5
-0.6
-6 .3
-5.8
-3 .1
0.7
4.7
6.9
9.8
10.2
10.2
9.2
8.3
6.5
5.2
6.4
1.8
4 .0
5.9
7.7
7.9
6.8
4.4
5.7
4 . 1
3.8
4.4
VI-27
steel plows, etc., was delivered to agriculture, along with a modest number
of trucks.
It has not been possible in this study to compile detailed estimates of
this capital, but a series for the value of tractors is presented in Table VI-15,
column 2. It will be seen that by 1934 the value of tractors on hand, in base-
period prices, had grown to exceed the value of the remaining horses. According
to Jasny's more comprehensive effort to assemble estimates for the sum of all
animal and mechanical power available in Soviet agriculture (displayed in the
graph above in chapter V, page V-6), the 1928 level of total draft power was
not regained until 1938. Though the timing of the changes is uncertain, it is
nevertheless clear that reduced draft power caused serious difficulties over
this period. As for other fixed assets, we have taken refuge in the crude
assumption that stocks remained constant from 1928 through 1940, i.e., that
neither destruction nor construction caused major changes in what was available.
We also make no effort to estimate the value of land as an input, following
established Soviet practice. A more detailed analysis of agriculture might well
distinguish a third subdivision for so-called technical crops (cotton, sugar beets,
oilseeds, etc.), but this too is put aside for another investigation. We are
left, then, with the two total asset series as devices for insertion into our
baseline solution to model the impact of collectivization.
Modifications to KAPROST for the Collectivization Scenario
Imposing on the model the actual levels of capital stock and output in
the two agricultural sectors will also fix investment in, and inter-industry
deliveries to, these sectors - exactly what we require. There are several ways
to incorporate these changes in KAPROST. Our choice was to add a new constraint
placing an upper limit on new technology capital stock, and then adjust capital
VI-28
Table VI-15
Actual fixed assets in fieldcrop and livestock-products agriculture, USSR, by-year, 1928-1940, in billions of rubles at base-priced prices.
Jan. 1
Fieldcrops Agriculture
Horses Tractors Other Total
Livestock-products Agriculture
Herds Other Total
1929
30
31
32
3
3
3
2
.8
.6
.1
.5
0.1
0.1
0.3
0.9
6.2
6.2
6.2
6.2
10.1
9.9
9.6
9.6
4.0
3.1
2.1
1.5
3.7
3.7
3.7
3.7
7.7
6.8
5.8
5.2
1933
34
35
36
2
1
1
1
.0
.8
.7
.8
1
2
4
6
.7
.8
.4
.3
6.2
6.2
6.2
6.2
9.9
10.8
12.3
14.3
1.3
1.5
2.0
2.9
3.7
3.7
3.7
3.7
5.0
5.2
5.7
6.6
1937
38
39
40
1
1
2
2
.8
.9
.0
.l(a)
7.9
7.2
6.7
6.0
6.2
6.2
6.2
6.2
15.9
15.3
14.9
14.3
3.0
3.3
3.6
3.7(a)
3.7
3.7
3.7
3.7
6.7
7.0
7.3
7.4
1941 2.1(a) 5.0 6.2 13.3 4.2(a) 3.7 7.9
(a) Including livestock on acquired territory.
Sources: The horse and non-horse livestock data are derived from estimates in Moorsteen
and Powell, The Soviet Capital Stock, 1928-1962, pp. 105, 107, and 337. The
tractor series is obtained by applying an assumed 5-year life to tractor pro-
duction figures for 1926-1940 in TsSU, Promyshlennosti SSSR, 1964, p. 279, and
linking the resulting stock series to a ruble value at the end of 1928 of 54
million rubles, in KTs 29/30, p. 446. Other assets are valued in 1929 as ex-
plained in Appendix A, and assumed unchanged thereafter.
VI-29
productivity so that if all possible capital stock is built and utilized,
just the historically observed output levels result. Note that this provides
the optimizing algorithm with the option of not building or not using some
portion of agricultural capital stock in any of the periods — with resulting
loss in output as final capital. In practice, however, agriculture is a
binding constraint on the solution and all capital allowed is built.
First we estimate the levels of capital stocks. As old technology
capital stock simply depreciates, and is never increased, its values can be
calculated for each period. New technology capital stock which is carried
over from each period (or which is already established) can be calculated re-
cursively by depreciating the previous years total new technology capital. The
sum of these two items, when subtracted from total capital, yield each period's
additions to new technology capital. Total new technology capital is the sum
of the established portion and additions, while the available new technology
capital counts only .7 of the additions in each period. These calculations are
presented in Table VI-16 for field crops and Table VI-17 for livestock. Note
that the reduction in herds is so great that it, strictly speaking, requires
negative increments to capital stock in the second and third periods. As the
MPS solution algorithm will not permit this type of outcome, we specify values
which lead to zero additions to capital by lowering old technology capital in
those two periods. This tends to underestimate the negative impact of collecti-
vization, as no investment is obviously less costly than positive investment
followed by destruction.
Given capital estimates, we adjust productivity in field crops by
assuming old technology capital productivity is constant, and altering new
technology productivity so that the available new technology capital will
produce the remainder. (See Table VI-18). For livestock, even after reducing
VI-30
Table VI-16
Capital Stock - Agriculture, Field Crops
Period
1
2
3
4
5
6
(1)
K°
7509
6259.3
5217.5
4349.2
3625.4
3153.1
In millions of rubles.
(1) Old technology capital
(2)Established
KN
2591
2215.5
2856.5
4003.9
6798.5
10495.6
(3)Total
K
10100
9600
9900
12300
15900
14900
stock, depreciated from
(A)
zN
-1125.2
1826
3946.9
5476.1
1251.3
1928 value
(5)
KN
2591
3340.7
4682.5
7950.8
12274.6
11746.9
(6)Available
KN
2591
3003.1
4134.7
6766.7
10631.8
11371.5
(2) New technology capital before including additions to capital for that period;depreciated value from previous period, column 5.
(3) Total capital stock, from Table VI-15, column 4.
(4) Additions to capital each period, = (3) -((1) + (2)).
(5) Total new technology capital stock = (2) + (4)
(6) New technology capital stock available = (2) + .7 x (4).
Period
1
2*
3*
4
5
6
(1)
K°
7011
5212.3(5975.5)
4498.6(5092.9)
4340.7
3699.6
3022
Table VI-17
Capital Stock - Agriculture,
(2)Established
KN
689
587.7
501.4
427.7
1159.5
2559.5
(3)Total
K
7700
5800
5000
5700
6700
7300
Livestock
(4)
zN
--
-
931.6
1840.9
1718.5
(5)
KN
689
587.7
501.4
1359.3
3000.4
4278
(6)Available
KN
689
587.7
501.4
1079.8
2448.1
3762.5
In millions of rubles(1) Old technology capital stock depreciated from 1928 value, except see * below.
(2) New technology capital before including additions to capital for that period; depre-ciated from the previous period, column 5.
(3) Total capital stock from Table VI-15, column 7.
(4) Additions to capital each period, = (3) -((1) + (2))).
(5) Total new technology capital = (2) + (4)
(6) New technology capital stock available for production = (2) + .7 x (4)* Value used differs from value obtained by depreciation (in parentheses) for reasons
explained in text.
VI-31
capital stock levels and forcing no additions for two periods, old technology-
capital at original productivity levels produces more than the observed output.
As a result, we summed old and available new technology capital, and divided
this total capital into total output to obtain a common productivity ratio
(Table Vl-19).
The Collectivization Solution
Adjusting the reference solution after applying these modifications
consisted of trying to raise consumption and lower unemployment as much as
possible within the new framework. The average propensity to consume was
lowered to .4 in all periods. The fraction of consumption required to be in
fixed proportions is raised in the fourth period to attain reasonable values
of optional consumption — too large a contribution from trade and distribution
arises otherwise — and lowered slightly in the last two periods to allow more
employment. These parameter changes are presented in Table VI-20.
The resulting solution appears in Table VI-21. Compared to the reference
solution, final capital stock actually increases to 358 billion rubles from 327
billion rubles. This is a reaction to the inability of the optimizing algorithm
to invest in agriculture or to produce output from sectors dependent on inputs
from agriculture, i.e., consumer goods. Producer-goods industry benefits. Of
course consumers pay the bill: consumption is down significantly in all periods
after the first; unemployment is up in all periods, and exceeds 25% in the last
four periods. It should be noted that this is not unemployment in the Western
sense, but indicates the presence of supernumerary laborers and a decline in
labor productivity throughout the economy. Finally, note that sixth period out-
put is down to 320 billion rubles from 406 in the reference solution, an indi-
cation that the lagging agricultural sectors are undermining the whole economy.
VI-32
Period
1
2
3
4
5
6
(1)
X°
18341.9
15289.2
12744.6
10623.6
8855.5
7381.2
Table VI-18
Productivity - Agriculture, Field
(2)Total
X
22752
23599
23126
24072
25393
26317
(3)
xN
4410.1
8309.8
10381.4
13448.4
16537.5
18935.8
Crops
(4)Available
2591
3003.1
4134.7
6766.7
10631.8
11371.5
(5)
6N
1.70209
2.76708
2.51080
1.98744
1.55548
1.66520
In millions of rubles
(1) Old technology output, obtained by depreciating 1928 values
(2) Total output, from Table V-9
(3) = (2) - (1)
(4) Available new technology capital stock, from Table VI-16, column 6
(5) = (3) ÷ (4), adjusted productivity for new technology capital.
Period
Table VI-19
Productivity, Livestock-Products, Agriculture
(1) (2) (3)Total Productive
X K
1
2
3
4
5
6
In millions of rubles
(1) Total output, from Table V-9.
(2) Total capital stock = Old technology capital from Table VI-17 column 1 plusavailable new technology capital from Table VI-17, column 6.
(3) = (l)÷(2), adjusted productivity for old and new technology capital
10320
7567
6172
8203
10320
10498
7700
5800
5000
5420.5
6147.7
6784.5
1.34026
1.34066
1.23440
1.51333
1.67868
1.54736
VI-3S
Table VI-20. Parameter Values: Collectivization versus
Reference Solution
Parameter
Average Propensityto Consume *
1929+19301931+19321933+19341935+19361937+19381939+1940
Fraction of Consumption inFixed Proportions *
1929+19301931+19321933+19341935+19361937+19381939+1940
CollectivizationScenario
0.40.40.40.40.40.4
0.10.10.50.70.850.85
ReferenceSolution
0.50.70.7
0.10.10.50.50.950.95
* These parameters are the same for both workers and peasants.
All other parameters and weights are the same for both solutions
VI-35
Trade, Defense and Collectivization
Combining all three influences to approach a composite reconstruction
proved much easier than we had expected. Starting with the combined trade and
defense solution, Table VI-13 above, the agricultural capital and output modi-
fications were added. The resulting solution had several problems: the sectoral
distribution of consumer demand was out of balance, final capital stock was very
low, unemployment was a bit high, and required imports were much too stringent
in the final period.
The necessary adjustments included lowering the required import ratios
to 15% of their initial level, from 20% in trade and defense solution. To
make consumption demand more reasonable, the fraction of consumption required
to be satisfied in fixed proportion was raised to .4 and .6 in periods three
and four (from .3 and .4) and lowered to .8 (from .9) in period six. All other
parameters, including the objective function weights, are identical with those
in the rearmament solution (table VI-11) and the defense and trade solution
(Table VI-13).
The composite reconstruction, combining trade, rearmament and collectiv-
ization is presented in Table VI-22. Note that here the capital stock figures
do not include military hardware. There are three general points to be made
regarding this solution. First, capital stock hold up very well. Second, the
sharp fall in consumption and rise in unemployment in period three appear to
be unavoidable; efforts to alter the model solution to alleviate this damage
were not successful for moderate costs in other objectives. Third, the very
small increment in capital stock in 1941-42 reflects the diversion of resources
away from capital formation and into defense procurement.
VI-37
Comparing; These Solutions with the Actual Record
How closely does our composite reconstruction, reflecting the combined
impacts of collectivization, rearmament, and world depression, correspond to
the actual historical record? In many respects we cannot be sure, since important
aspects of the record remain uncertain. But some overall checks are available,
and they indicate that our composite reconstruction matches up fairly closely
with the historical record. For example, comparison of the economy-wide gross
output totals in our composite reconstruction with those compiled above in
chapter V and displayed in Table V-9 demonstrates that our estimates clearly
have the right dimensions. Data for the six periods appear in the first two
lines of the gross output panel in Table VI-23, where it will be seen that our
composite reconstruction starts out slightly below the historical level but
catches up and surpasses it.
This higher growth rate can be traced to two influences. First, the
optimizing algorithum has certain advantages that planners at the time were
unlikely to have had. This supra-normal efficiency we would expect to lead to a
higher growth rate. But it also leads to slack capacity in some sectors in the
first two periods; unaffected by the inertia which haunts real economic systems,
KAPROST does not produce anything it cannot use! Second, as noted above, many
of the negative impacts on the Soviet economy - the purges and terror, the
transportation bottleneck, the initial overinvestment and subsequent waste
were not quantifiable for inclusion in KAPROST. While some of these may have
been picked up in our data (the agricultural output and capital stocks, for
instance), failure to capture them fully will lead to better performance in the
model than in real life.
VI-38
A second confirmation of our composite reconstruction comes from the
correspondence between its capital stock series and the most reliable available
estimate of Societ capital stocks. We show in the first line of the panel
on capital stocks in Table VT-23 the index for gross capital in 1928 prices
compiled by Moorsteen and Powell (The Soviet Capital Stock, 1928-1962, Table
T-28, p. 322), chained to our 1928 ruble figure. The composite reconstruction
tracks the Moorsteen-Powell index closely, though again our estimates are
somewhat below the actual record in the early years and somewhat above it at
the end.
The drastic fall in consumption displayed by our composite reconstruction
from 38 billion rubles in 1929 and 1930 to 32 billion in the next two years and
only 21 billion in 1933 and 1934, cannot be checked against official or Western
estimates, since no continuous estimates are available. Professor Chapman long
ago demonstrated conclusively, however, that the non-agricultural real wage in
1937, after substantial recovery from the most difficult years around 1933, was
17% below its 1928 level (in terms of the 1928 price structure), or 43% below
its 1928 level (in terms of the 1937 price structure). Cf. Janet G. Chapman,
Real Wages in Soviet Russia Since 1928 (1963) , pp. 145-46.
Perhaps interactions within the composite reconstruction that push its
first period total consumption about the 34.4 billion rubles of the reference
solution should be discounted as unrealistic; perhaps even the rise of the
reference solution figure from 30.5 billion in the base period (twice in 1928
level) should be treated as unlikely. These increases reflect KAPROST's
dexterity in distributing output. As capital investment is limited by capacity
bottlenecks in the construction sector, extra output from the agricultural,
housing and consumer goods sectors is delivered to consumers. A more sophisticated
Table Vl-23. Comparing Solutions — Results for Gross Output and Total Capital Stocks,
By Period, in Billions of Rubles at Base Period PricesP E R I O D
GROSS OUTPUT
Actual (Table V-9)
Composite (Table VI-22)
Agricultural Collectivization (Table VI-21
Depression + Rearmament (Table VI-13)
Rearmament (Table VI-11)
World Depression (Table VI-5)
Reference Solution Table IV-19)
TOTAL CAPITAL STOCK, AT START OF YEAR
Actual
Composite
Excluding defense capitalIncluding defense capital
Agricultural Collectivization
Depression + RearmamentExcluding defense capitalIncluding defense capital
RearmamentExcluding defense capitalIncluding defense capital
World Depression
Reference Solution
1929+1930
127
115
)117
119
121
121
122
1929
67.3
67.167.1
1931+1932
151
139
140
148
149
149
149
1931
80.5
71.871.8
1933+1934
169
154
161
171
167
182
182
1933
100.9
93.193.1
1935+1936
231
236
244
250
252
257
258
1935
123.5
125.3128.3
1937+1938
281
307
311
331
340
330
332
1937
160.2
164.6184.3
1939+1940
307
324
320
402
419
407
406
1939
202.6
239.1287.9
1941
265.9
229.1337.9
67.1 71.8 95.7 123.9 183.3 273.0 358.1
65.765.7
65.765.7
65.7
65.7
74.674.6
74.674.6
74.6
74.6
98.098.0
96.296.2
102.4
102.5
125.5128.5
127.0130.0
121.8
122.1
171.1190.8
174.5194.2
191.4
192.5
244.7293.5
250.7299.5
244.9
248.1
262.3371.1
276.4385.2
324.2
327.4
Table VI-24. Comparing Solutions — Results for Total Consumption and Unemployment,By Period, in Billions of Rubles at Base-Period Prices, and Percent Unemployed
P E R I O D
TOTAL CONSUMPTION
Composite (Table VI-22)
Agricultural Collectivization (Table VI-21) 36.4
Depression + Rearmament (Table VI-13)
Rearmament (Table VI-11)
World Depression (Table VI-5)
Reference Solution (Table IV-19)
UNEMPLOYMENT (PERCENT)
Composite
Agricultural Collectivization
Depression + Rearmament
Rearmament
World Depression
Reference Solution
1929+1930
37.7
36.4
36.1
36.2
34.7
34.4
1931+1932
31.7
30.0
35.6
36.2
33.4
33.1
1933+1934
20.9
25.8
32.3
27.5
40.4
41.0
1935+1936
27.3
29.6
31.1
30.8
40.6
40.6
1937+1933
32.3
31.7
37.8
38.4
66.6
66.7
1939+1940
36.3
35.6
47.1
48.4
82.6
82.3
13.9
12.4
9.3
8.7
8.5
7.8
21.5
19.6
13.1
12.4
12.0
11.6
34.5
28.6
14.2
23.2
0.7
0.0
26.6
24.3
4.9
7.7
0.0
0.0
26.8
27.9
7.4
7.0
0.0
0.0
29.6
31.3
0.0
0.0
0.0
0.0
VI-41
model with inventory behavior might have held back some of this for use in
later periods, especially the producer goods. Alternatively, a less efficient
economic system might have lost some of it through waste.
In any case, the fall from 30.5 billion in the late 1920's to 20.9 billion
in 1933 and 1934 represents a 31.5% shrinkage in real consumption. If the 20.9
billion is compared with the third period level of 41.0 billion attained under
the problem-free conditions of the reference solution, the implication is a
49% shortfall in comparison to what might have been.
Another tragic feature of the composite reconstruction is the massive
volume of unemployed labor it involves. The unemployment is organically linked
to the shrinkage in real consumption. Since we have maintained the same physical
labor productivities throughout all the tests from the reference solution on,
the successive constrictions on the economy have made more and more labor
superfluous. The unemployment panel on Table VI-24 shows the results. By
comparison with the reference solution, which made use of all available labor
after the structural shifts of the first two periods were completed, rearmament
altered resource use in a way that generated unemployment in periods three, four,
and five as well. But the impact of agricultural collectivization was far more
severe, making between 24% and 31% of the total labor force unnecessary.
Combining collectivization with world depression and rearmament, our composite
reconstruction generates unemployment levels running from 14% in the first
period to 34.5% in the third, 27% in the fourth and fifth, and 30% at the end of
the decade.
Since by constitutional definition unemployment did not exist in the
USSR, government policy kept the whole labor force at work: on payrolls in
VI-42
non-agricultural occupations or in sovkhozy, as members of kolkhozy, or in
corrective labor camps. Average labor productivity per person in physical terms
fell far enough below its potential to absorb the superfluous labor. Massive
unemployment was thus camouflaged, by contrast with the explicitly exposed
unemployment in the West during the same period.
The results displayed in Tables VI-23 and VI-24 reveal instructive
contrasts. Beginning at the bottom of each panel, one can proceed upward from
the reference solution - our hypothetical best case - step by step to the
composite reconstruction, noting which factors had the strongest impact.
Inspection shows very clearly that agriculatural collectivization dealt the
most serious blow against the economy, far more crippling than the two external
factors of rearmament and world depression. The evidence is especially vivid
in terms of household consumption. Total household consumption over these
twelve years comes to 298 billion rubles in the reference solution; world
depression leaves the total unchanged. Rearmament lowers total consumption by
27%, to 218 billion, and the two exogenous factors in combination have the
effect of restricting consumption to 220 billion. But collectivization of
agriculture reduces the total by 37%, to 189 billion, and our composite
reconstruction shows a twelve-year total for household consumption of 186
billion rubles.
The loss of draft animals and productive livestock did far more than
reduce the flow of livestock products to the population. It forced industry to
deliver tractors and other agricultural machinery to field-crops agriculture
more promptly than had been intended, thus making non-agricultural capital
formation more difficult. It lowered labor productivity in agriculture directly,
and indirectly throughout the economy as well. One sees this in Table VI-23.
VI-43
where, though the total capital stock does not seem severly reduced by the
imposition of collectivization, gross output in 1939 and 1940 is 25% lower in
solutions including collectivization than in any of the others. The capital has
been delivered in less productive uses, and the shift is obtained also at the cost
of severly reduced consumption.
Probably the most important consequence of collectivization, though it
is not directly measurable by KAPROST, was the political tension and pressure
that this "second revolution" spread throughout Soviet society, culminating in
political purges and mass terror. This is the tragic human story that underlies
the laconic figures in our results.
To facilitate comparisons between our composite reconstruction, which is
meant to approximate what actually happened, and the reference solution set forth
at the end of chapter IV, we offer some overall data on GNP components and shares
in Table VI-25. Several striking contrasts are evident. The total GNP in our
composite reconstruction is smaller than in the reference solution over all six
periods, as is only to be expected. The shortfall is modest at first, but it is
15% in the 4th period, 11% in the 5th period, and 21% in the 6th period, as the
negative developments impose their combined effects. Consumption suffers markedly,
as we have seen, but these sacrifices do not serve to make investment larger
in the composite reconstruction than in the reference solution. Collectivization
and rearmament combine to hold investment down, especially in the last two years,
when procurement rises rapidly and fixed capital formation falls from 70 to 21
billion rubles.
The share of consumption in GNP declines from its starting point in both
solutions, as resources are shifted into capital formation. But the belt-
tightening is more severe in the composite reconstruction, where consumption
VI-44
Table VI -25 . Comparison of GNP and I t s Components in the CompositeReconstruction and in the Reference Solution
P E R I O D1929+ 1931+ 1933+ 1935+ 1937+ 1939+1930 1932 1934 1936 1938 1940
BILLIONS OF RUBLESConsumption
Composite ReconstructionReference Solution
InvestmentComposite ReconstructionReference Solution
Defense ProcurementComposite ReconstructionReference Solution
Government Services
37.734.4
15.320.0
31.7.33.2
31.133.7
20.941.0
42.941.8
27.340.6
59.779.5
32.366.7
69.979.4
36.382.4
20.797.2
3.0 17.0 31.0
PERCENT SHARESConsumption
Composite ReconstructionReference Solution
InvestmentComposite ReconstructionReference Solution
Defense ProcurementComposite Reconstruction
Government ServicesComposite ReconstructionReference Solution
ExportsComposite ReconstructionReference Solution
0.0 0.0 3.5 13.1 18.5
64.9
CompositeReference
ExportsCompositeReference
ReconstructionSolution
ReconstructionSolution
Gross National ProductComposite ReconstructionReference Solution
10.97.7
2.01.7
62.665.2
12.110.1
2.32.3
77.081.1
14.712.5
1.93.0
86.2101.9
12.514.9
1.53.6
129.4145.6
12.017.3
1.34.2
167.3178.7
22.719.7
1.14.8
171.9218.8
60.252.8
24.430.7
41.240.9
40.441.6
2440
4941
.2
.2
.8
.0
21.127.9
46.154.6
1937
4144
.3
.3
.8
.4
21.137.7
12.044.4
37.8
10.911.8
3.22.6
12.112.5
3.02.8
14.712.3
2.22.9
12.510.2
1.22.5
12.09.7
0.72.3
22.79.0
0.62.2
VI-45
accounts for only one-fifth of the GNP in the last three periods. This extra-
ordinarily low share excludes public consumption, i.e., health care, education,
and municipal services, which account for from 11% to 23% of the GNP, so it is
not quite so shocking. Nevertheless it is clear that the Soviet people bore
a very heavy burden during this period. Half the GNP was channeled into accum-
ulation in 1933 and 1934 under the composite reconstruction, but the investment
share fell off to 42% in the fifth period and dropped to 12% in the sixth, as
defense procurement rose to account for more than a third of the GNP in 1939
and 1940.
For those who may wish to examine further details, sector by sector and
period by period, we conclude this chapter by reproducing the five reports
generated by KAPROST for the composite reconstruction. These are the specific,
concrete estimates of output and resource use that should be compared with
available independent evidence in appraising the accuracy of the composite
reconstruction.
Appendix A Statistical Foundations
Pages
A. Dimensions and coverage of the data 3-6
1. Prices2. Dates3. Sectors
B. The 1928 and 1933 flow tables 7 - 2 3
1. Agriculture: field crops and livestock products2. Industry: producer goods and consumer goods3. Electric power4. Transport and communications5. Construction6. Housing7. Trade and distribution3, Government services9. Deliveries to fixed capital
C. Capital stocks 2 3 - 2 7
1. Unadjusted data2. Separating out electric power and construction3. Deducting unfinished construction4. Depreciation
D. Population and labor force 27-341. Population aged 15-592. Sectoral employment, wage rates, and wage bills3. Consumption/income ratios
E. Foreign trade 35-37
1. Exports2. Required and optional imports3. Trade debts
-2-
Appendix table3
Page
1. Agricultural gross output, 1927/28, 1928, and 1932/33, alternativeversions, in millions of rubles. 8
2. Gross output of individual field crops and livestock products, 1928and Plan 1933, in millions of rubles at 1926/27 prices. 10
3. Gross output of census and small-scale industry, 1927/28, alternativereports, in millions of rubles at 1926/27 prices. 13
4. Summary national balance for 1928, as compiled in HP.332, pp. 184-207,in millions of rubles at 1928 prices. 14
5. Transport sector revenues, 1928 and 1933, by mode, in millions of rubles.19
6. Stocks of total fixed capital at the start of 1928, 1929, and 1933, bysector, in millions of rubles at 1925/26 prices. 25
stocks7. Adjusted^of completed and unfinished capital, by 3ector, in millionsof rubles at 1925/26 prices. 27
8. Projected additions to and subtractions from the population aged 16-59,by year, 1928-41, in thousands of persons. 29
9. Hypothetical population aged 16-59 and engaged labor force, by year, 311928-41, in thousands of persons.
10. Annual wage, labor force, and wage bill, by sector, 1928 and 1933. 33
-3-
Appendix A. Statistical Foundations
Dimensions and coverage of the data
Conceptually this study attempts to present its material in real terms,
but its aggregate data are necessarily ruble magnitudes. As much as
possible, they employ base period prices, usually in the form of "1926/27
rubles." Official Soviet statistics using "1926/27 rubles" became increas-
ingly dubious after 1930 or so, as Professor Bergson and many others have
shown. In this study, however, they are only used as measures of the base-
period situation in 1923, or as indicators of expansion intentions expressed
in the first FYP. Soviet prices in the late 1920s may have undervalued
fixed capital relative to current output, and underpriced manufactured goods
relative to primary products, but the general level only rose by two or
three percent from 1926/27 to 1927/28, and the "1926/27 rubles" seem to
provide an adequate set of weights for our purposes. Many targets for the
terminal year of the first FYP were estimated by Gosplan, not only in terms
of "1926/27 prices," but also in terms of values that would reflect sub-
stantial reductions in construction costs and production costs. These
targets in "reduced cost" prices understate the intended physical change in
outputs and inputs, mixing price and quantity changes in ways that would
confound orderly analysis. They are therefore ignored in this study.
In forward-looking planning, ex ante prices have to be employed. One
might conceive of a set of relative prices emerging as the dual to a
terminal-year solution for a vast linear programming problem incorporating
all intervening real structural changes in the economy, but the concept
itself was not available fifty years ago, nor would a sufficiently disaggre-
gated solution be readily obtainable even today. Application of late-year
price weights, reflecting the impact of structural change, is certainly
relevant in measuring growth rates, and if this project were on a larger
scale it could investigate the impact on measured structural change of late-
year weights. Constrained by resources, however, and since our purpose is
to investigate the Soviet economy's potential for producing a new, intended
bill of goods, we can take some analytic consolation from the late Richard
Moorsteen's finding that base-period weights are conceptually the most
appropriate weight3 for aggregation.
Most of the ruble data drawn on for this study were originally in
purchasers prices, including trade markups, freight charges, and excise
taxes, but other evidence, especially that contained in the Materialy po
balansu narodnogo khoziaistva SSR, issued by the Chief Administration for
K
National Accounts in October 1932, makes it possible to derive flow-table
estimates in producers prices for the rows of an input-output table. The
details are set forth below. As for data on capital stocks, Gosplan cots-
piled estimates in "1925/26 prices" for the first FYP, followed by estimates
in almost identical coverage in "1926/27 prices" for the 1929/30 Control
Figures. The national total for October 1, 1928, was put at 70,154 million
rubles in "1925/26 prices" and .69,483 million in "1926/27 prices," a
difference of less than one percent. As with output, the terminal-year, s
target for capital stocks employed here are those assuming no change in
prices, rather than those anticipating marked reductions in construction
costs.
The fiscal year in use at the start of the first FYP covered a twelve-
month period from October 1 through the following September 30. The base
year preceding the plan period therefore ran from October 1, 1927, through
- 5 -
September 1928; the terminal year ended September 30, 1933. In this study,
these two years are referred to as 1928 and 1933. During the plan period,
planning and administration were shifted to a fiscal year coinciding with
the calendar year, beginning with January 1, 1931; the "special quarter" at
the end of 1930 was made the occasion for extra campaigns for over-fulfillment
of plan targets. The shift does not, however, affect the computations in
this project.
The capital stocks on hand and intended are set forth in the FYP by year
and by sector, together with gross investment and annual depreciation estimates,
for the three pre-plan years 1925/26-1927/28 and for the full plan period,
1923/29-19 32/33. Following Eckaus and Parikh, I take the capital stocks on
hand at the beginning of a year as the constraints on that year's activity.
For base period output/capital ratios, therefore, I use active October 1,
1927 capital stocks in the denominator, and base year output levels in the
numerator for each sector. For terminal-year output/capital ratios, I deduce
the October 1, 1932 active stocks implied by growth toward the October 1, 1933
targets and compare them with intended terminal-year output levels. One could
center the capital stocks at midyear for elegance, but the impact on computa-
tions would be negligible.
Our purpose here is to trace large flows among the major sectors of the
economy, including the income and product flows of the usual national accounts, -
but covering also the entries in the northwest quadrant of an economy—wide
input-output table. Soviet research in this direction had already led to
publication of a set of national balances for 1923/24, and fragments of
similar elements and computations appear in the 1929-30 Control Figures and
-6-
in Che first FYP. The Soviet statistical concept of gross output is roughly
similar to gross output in the input-output sense. Plan data are incomplete
at various points, however, so substantial rearrangement and amplification
is required in order to build the two crude tables employed in this study.
The agricultural sector here is divided into field crops and livestock
products; forestry is included with field crops while fishing and hunting
are included with livestock products. Until the mid-1930s, Gosplan included
the construction industry, i.e., the erection of new capital plant and in-
stallation of new capital equipment, in the industrial sector, along with
the industry manufacturing construction materials. For this study I have
attempted to subtract out the fixed assets and current output of the construc-
tion sector to form a separate sector whose capital-forming activities play
a crucial role in the model. Gosplan documents, on the other hand, devote
substantial attention to electrification, and show the electric power
industry separately from other industry. Given its high priority, it is
thus informative to retain electric power as a separate industry. All
remaining industry is divided between producer goods and consumer goods as
explained below. Plan data on the communications sector are here combined
with data for the transportation sector to make up a single sector. Plan
data on capital stocks show very substantial amounts cf urban and rural resi-
dential capital, so a housing sector clearly requires recognition. The flow
of services from this capital must be crudely estimated, however, since Soviet
plan accounts do not recognize this form of output. Gosplan data on assets
make it possible to distinguish two other sectors, one for trade and distri-
bution, and another for government services: education, public health,
municipal services, and government administration. Output measures for these
two sectors are less certain, but the available clues are utilized.
— 7 —
The 1928 and 1933 flow tables
Table 1 presents summary data for several main categories in gross
agricultural output, illustrating the kind of coverage problem that is in-
volved in constructing flow tables for the Soviet economy. The first three
columns related to fiscal 1927/28; the fourth column to calendar 1928; and
the last two columns to the terminal year of the first five-year plan,
figures are reported in "1926/27 prices," except that the fourth column
figures are in 1928 current prices. A glance at the table discloses modest
differences in figures that appear to represent the same item, together with
several gaps. Column six shows the extent to which agricultural targets in
the "optimal variant" were more ambitious than the "basic variant" targest
(in column 5) for agriculture in the first FYP.
The central agricultural activities of growing field crops and producing
livestock products are surrounded by several peripheral activities whose
dimensions have to be deduced from inclusions and exclusions in Soviet sources.
In the FYP, for example, the all-inclusive totals for columns 2, 5, and 6
appear on pp. 162-65 of Vol. I, together with a smaller figure for zemledelie
("agriculture pro p e r " ) . Data for forestry appear in FYP, Vol. I I , part 1,
p, 372, so a residual I take to be for fishing and hunting is precipitated.
For 1927/28 it checks precisely with the fishing and hunting output given in
the 1928/29 control figures (Kontrol'nye tsifry nar, khoz. SSSR na 1923/29
god — hereafter KTs 28/29), p. 398. Further detail appears in KTs 28/29 at
p. 476, and, as entered in column 3, in the control figures for 1929/30
(KTs 29/30), at pp. 534-35 and 422-24. These make the entries in FYP, Vol. II,
part 1, pp. 334-35 more identifiable. All purport to be in 1926/27 producers'
prices, and the generally small variations are not further explained. The
column 4 figures are for calendar 1928 in 1928 prices, from Materialy no
. 8 ,
Table 1. Agricultural gross output, 1927/28, 1928, and Plan 1932/33,
alternative versions, in millions of rubles.
1928/29 First 1929/30control Five-year controlfigures Plan figures
1932balan-
ces
1933basic
targets
1933optimaltargets
Field crops;
All grainsVegetablesTechnical cropsMeadow grassForestry
Total field cropsand forestry
33
11
2
739751901953765
109)
33
11
(10
731032911542765
981)
34
21
(12
551596965105757
974)
42121
(11
449.4842.7046.6072.2411.3
822.2)
54112
(15
2 32649619912596
823)
55112
(17
618085854912596
065
Livestock products:
Meat, poultry, milkFish and gameRaw materialsGrowth of herdsManure
Total livestock products,hunting, and fishing
Total for agriculture,hunting, fishing, andforestry
114368545266244
NA368NA546NA
4 438391642-79350
4 449.2323.5623.2NANA
NA580NA
1 002NA
NA580NA
1 002NA
(5 537) (5 673)
17 646 16 659
(5 742)
18 716
(9 153.9
20 976.1
(8 050)
23 878
(8 741)
25 806
- 9 -
balansu nar. khoz. SSSR za 1923, 1929, i 19 30 g.g. — hereafter MPB, pp. 139,
109, and 185.
The figures in Table 2 provide greater detail on the items making up
the entries in Table 1. They reflect application of average annual 1925/27
prices (from KTs 29/30, pp. 581-32) to physical quantities. The 1923
quantities are drawn from MPB appendix tables, pp. 312-31 and 280-85 (for
timber and firevood). The "optimal variant" targets for 1932/33 are from
the FYP, The computed ruble amounts, in turn, are generally quite close to
those reported for 1927/28 KTs 28/29 and KTs 29/30 if they appear there.
The compilers of MPB assembled remarkable detail on the sources and
uses of all these items, showing physical amounts, ruble amounts, and quo-
tients (average prices) for all but a few cells, with the price varying,
cell by cell, across each row, I multiplied the physical quantities by the
average 1926/27 producer price of each gross output instead, and sunned the
resulting ruble values delivered to each receiving sector in order to obtain
flow-table rows for field crops (including timber and firewood) and for field
crops (including timber and firewood) and for livestock products (including
fish and game, but excluding growth of herds).
No such detail being available in the FYP, I applied the 1928 row struc-
ture to the 1932/33 gross output targets for each crop or product to obtain
an estimate of the anticipated flows to receiving sectors. Since the in-
tended increases of 1932/33 over 1927/28 varied from product to product —
see column 3 in Table 2 •— the composite structure of field-crop deliveries
and livestock product deliveries implied by the plan targets differed sorne-
what from the base-period structure.
-10-
Table 2. Gross output of individual field crops and livestock
products, 1928 and Plan 1933, in million of rubles at 1926/27 price3
All livestock products 5 301 8 185 5b
Sources: For 1928, see text. FYP, II, 1, pp. 338-39 for 1933 targets,
and KTs 29/30, pp. 581-82, for prices, except as noted, (a) FTP, II,
1, P. 335. (b) FYP total nongrain nontechnical minus above estimates for
potatoes and hay. (c) FYP} II, 1, p. 372. (d) FTP, I, pp. 162-65.
(e) 1928 figure raised in proportion to intended growth of herds.
(f) Totals less enumerated items.
-11-
Summary data on utilization of the grain crop, actual in 1927/23 and
intended in 1932/33, appear in a "bread-forage balance" in FTP, II, 1,
page 341. The amounts shown for livestock feed are treated in this project
as deliveries from fieldcrop agriculture to livestock-products agriculture,
except for the feed going to horses, which is an intrasectoral flow. For
the latter I have drawn on Naum Jasny, who analyzed grain crop utilization
in some detail in his Socialized Agriculture o_f_ the USSR (1949) , Note J,
pp. 748-60, supplemented by his paper, "Soviet Grain Crops and Their
Distribution," International Affairs, Oct. 1952, pp. 452-59. He shows annual
feeding rates for adult horses of 464, 593, 458, and 408 kilograms per head
from 1925/26 through 1928/29 on p. 753, and suggests an average for the
period of "about 450 kilograms" (p. 453), evidently considering 1926/27
atypical. Multiplying 21.4 million head of working horses (FYP, II, 1, p. 38)
by 458 yields 9.8 million tons of grain, 40.21% of the total used for feed
in 1927/28; the same share is assumed for 1933.
In connection with actual developments after 1928, one might assume
lower feeding rates under the stress of collectivization except that Jasny
notes (p. 458) "...a counteracting factor is the fact that large farms
generally tend to use more concentrated feed for their animals than small
farms...," and his further observation (p. 755) that "...the great shortage
of all kinds of draft power would have necessitated more work and more con-
centrated feed per work animal." Applying the same feeding rate and noting
the fall in the total number of horses from 32.6 million on Jan. 1, 1929 to
14.9 million on Jan. 1, 1935, suggests a proportionate fall in grain consumed
by working horses, thus releasing as much as 5.6 million tons of grain
annually for human consumption.
- 1 2 -
Five-Year Plan documents provide abundant data relating to industry
but, a s . with -agriculture, many i3sues of definition and coverage complicate
an effort to prepare unambiguous estimates for flow tables. The most de-
tailed stat is t ics came to Gosplan and TsUNKhU from so-called "census industry,"
especially those enterprises whose activities were planned by Vesenkha, There
were, however, enterprises in census industry not under the supervision of
Vesenkha, and there was a good deal of activity in so-called "small scale
industry." Table 3 records some gross output totals from our four principal
sources showing the diversity of summary data. Output figures are further
complicated by a distinction between gross output and net output (literally
translated as "the trade part"), which differed appreciably from source to
source , product to product to product, and year to year. Finally, industry
in the USSR has long been thought of as producing either producer goods
(Group A) or consumer goods (Group B), a distinction associated with the con-
trast between "heavy industry" and "light industry." Responsible Soviet
economists and statisticians readily recognized that some consumer goods were
produced in heavy industrial enterprises, and that some of the products of
light industry were consumed as intermediate inputs by industry, thus becoming
in fact producer's goods. They found in practice that neither classification—
by type of product or by type of producer — can be imposed and maintained
without overlap, ambiguity, and compromise.
The national balances released in 1932 employed a conceptual framework
providing a great deal of the structural information required for a present-
day input-output table. Within this framework a substantial volume of
detail is presented in appendix tables, extremely revealing for developments
in 1928, 1929, 1930, and (to some extent) 1931. Table A shows the main
categories appearing on both the sources and uses side of the annual balances;
-13-
Table 3. Gross Output of Census and Small-scale Industry, 1927/23,
Alternative Reports, in millions of rubles at 1926/27 prices.
Census industryPlanned by VSNKhOther
Small-scale IndustryAll industryExcise taxes
(1)1928/29 KTs
13 333
2 69516 5281 505
1310
23
161
(2)FYP
382909973034916396
(3)1929/30 KTs
14 13111 0673 0644 962
19 093
(4)(3) ad1.
14 11511 0673 0483 412
17 5271 402
(5)1932 Bal
19 245.0
5 868.725 113.71 463.0
Total incl . excise taxes 18 033 18 312 18 929
Sources: KTs 1928/29, pp. 400, 402; FTP, Vol. I I , pt. 2, pp. 84-85; KTs 1929/30,pp. 429, 592; MPB (1932), p. 185.
Note: The 1929/30 KTs data include flour millions, Col. 4 deducts estimates for flour-milling given in FYP, Vol. I I , pt. 2, pp. 120-21.
detailed entries for component elements are placed in the same format. At
the bottom of Table 4 is an indication of the adjustments decided on for use
in the present study. Losses (line 12) are subtracted from both sides of
the balance. Supplies on hand at the beginning of the year are. subtracted
from supplies on hand at the end of the year to obtain a measure of deliveries
to inventory. Total deliveries of domestic production (net of losses) to-
gether with imports are then traced as going to interindustry receivers (line 8)
and to the usual final demand claimants: households, government, investment,
exports, and inventory. The MPB s ta t i s t ic ians were meticulous in estimating
the incidence of trade-transport markups, excise taxe3, and import duties;
as a result deliveries in purchasers prices can be reduced to deliveries
in producers prices with fair precision. The MPB s ta t i s t ic ians placed hunting
and fishing with agriculture but placed forestry with industry; detailed data
on roundwood and firewood enabled me to shift forestry over to join the field-
crops part of agriculture. Following established Marxist-Leninist theory
-14-
Table 4: Summary national balance for 1928 as compiled in MPB 32, pp. 184-207, in
millions of rubles at 1928 prices.
Sources
1 Total supplies atbeginning of year
2 Annual productionat producers prices
3 Trade-transport markings
4 Excise taxes
5 Imports
6 Import duties
7 Total available atpurchasers prices
Uses
8 Total consumedin production
9 Total consumed bypopulation and institutions
10 Total delivered forfixed investment
11 Exports
12 Losses
13 Total supplies at endof year
14 Total distributed inpurchasers prices
15. Total supplies minusstarting supplies (13-1)
Total deliveries at purchasersprices (8+9+10+11+15)
Deductions (3+4+6)
Total deliveries at producers
prices (2+5-12)
Products Products of Products of Wholeof industry agriculture construction economy
-15-
the MPB statisticians did not, however, provide much information on the
output of nonmaterial economic activities; these had to be estimated in
ways explained below.
After a good deal of frustrating experiment i t has seemed best for
present purposes to make no use of Soviet figures for net output, relying
instead on the rather detailed appendix material in the 1932 MPB providing
breakdowns between "means of production" and "objects of consumption" as
they were , delivered to major sectors of the economy. Using these two cate-
gories, two rows can be developed for industry. Because of i ts high
priority, the electric power subsector of heavy industry is here separated
out from the producer goods row. At the same time i t is necessary to draw
as much as possible from the FYP itself for gross output figures relating
to actual 1927/28 and to intended 1932/33, since they provide the most con-
clusive evidence on the planners' own figures and intentions. These are
conveniently summarized in a table in FYP, I I , 2, pp. 78-79, shoving effective
demand for industrial goods, in millions of rubles at purchasers prices:
Demand for means of production:Agriculture and forestryIndustry and electrificationTransportCommunicationsTradeCommunal economyHousing construction"Nonproductive" commissariats
Total demand for means of production
Demand for means of consumption:Urban populationRural populationCollective consumption, by "nonproduc-tive" commissariats and other pro-organizations
Total demand for means of consumption
1927/28
7066 37292480290178460125
basicvariant1932/33targets
1 85210 3931 876103388350
866341
optimalvariant1932/33targets
1 95810 9002 280108481514
985408
9 135 16 169 17 634
33
1_
7
062318
010
390
45
1
11
474077
520
0.1
45
1
11
866441
617
924
- 1 6 -
Exports:Means of production 330 345 1 005
Means of consumption 76 159 167
The MP3 appendix tabula t ions permit subdivision of the de l ive r i e s to
agr icu l tu re and forest ry into four c e l l values in each year , ident i fy ing
the de l ive r ies to fieldcrop agr icu l tu re and l ives tock ag r i cu l tu re of both
producer goods and consumer goods (since MP3 records some consumer-good
de l ive r i e s not dis t inguished by the FYP).
The industry and e l e c t r i f i c a t i o n numbers require d i f fe ren t treatment
since MPB went considerably beyond Gosplan in est imating the in t e r indus t ry
de l ive r i e s that are very large and important. The MP3 tabula t ions separate
a l l industry in to census and smal l -scale indus t ry , and for each of then MPB
shows the i r gross output of means-of-production products and means-of-con-
sumption products . The tabula t ions also show the input rece ip ts of means-of-
production products by census industry and by small scale indus t ry , as follows,
in mil l ions of rubles at 1928 p r i c e s :
Produced Received
means of production 12 211,3 9 562.8
Census industry means of consumption 7 020.6 1 376.0
combined t o t a l 19 231.9 10 938.8
means of production 1 115.5 977.7
Small-scale industry means of consumption 4 667.9 536.9
combined total 5 783.4 1 514.6
Applying the crude standard assumption that census industry used i ts inputs
in the sane proportions for producing means-of-production products as i t did
for producing means-of-consumption products, one obtains estimates for the
flows of each input category into each output category; similar treatment yields
-17-
flows in small-scale industry as well. Adding census industry and small
scale industry together yields the following deliveries:
toproducer
goods
toconsumergoods
From producer goods 6 261 4 280
From consumer goods 977 936
These total 12,454 million rubles, almost twice the 6,372 million Gosplan
shows for 1927/28, but the added amounts appear plausible as intrasectoral
flows and at least crude cell values are needed for the interindustry tech-
nology matrix. A cell value for electric power is derived below.
For transport and communications, MPB estimates provided a percent
division between producer and consumer goods that is applied to the FYP? fig-
ures. For the construction sector MPB estimates are used instead of the
FYP figure which is limited to housing construction alone. The FYP figure
for "communal economy" is accepted as applying to producer goods used in the
sector supplying housing services to the urban population. For trade, the
FYP? figure is divided in MP3 proportions. The two FYP figure3 for "unpro-
ductive commissariats" are taken as applying to the government services
sector. The urban and rural consumption figures are subdivided in MPB
proportions since here again the MPB statisticians had more thorough break-
downs. The FYP export data are accepted as given. An estimate for additions
to stock is derived from year-end figures for 1926/27 and 1927/28, sub-
divided in MPB proportions. Finally, since all these figures are for domestic
production plus imports, measured in purchasers prices including a trade-
transport markup and excise taxes, they have been reduced across each row
by uniform percentages derived from the detailed MPB appendix information.
-18-
Industry rows for the 1933 flow table are built on the gross output
targets for producer-goods and consumer-goods production, assigned in the
ways intended by Gosplan, but subdivided in the proportions uncovered for
1928 by the M?3 s ta t is t ic ians . This approach fails to capture some of the
structural change that lay within the aggregates published in the first FYP
but at least i t reflects the diverse growth intentions planned for combined
deliveries to individual receiving sectors.
Deliveries of electric power can be deducted from the producer goods
row on the basis of scattered information in the first FYP and elsehwere.
The MPB study presents an electric power balance for 1928 in kilowatt-hours
and in rubles (p. 267); i t s KWH figures conform closely to those in FYP, II ,
1, p. 66, as indicated in the following tabulation:
1928MPB FYP 1933 basic 1933 optimal
Receiving sector:
Agriculture
Census industry
Transport and communications
Communal and consumption uses
Station use and transmission losses
Total 5 003 5 050 17 000 22 000
The 1933 figures in parentheses distribute the unstated optimal residual in
basic-variant proportions. The deliveries to census industry can be divided
into deliveries for means-of-production industry and means-of-consumption
industry on the basis of detailed absolute estimates for 1927/28 in the
TsUNKhU s ta t i s t ica l handbook, Narodnoe khoz. SSR: stat . sprav. 1932, pp. 40-41.
The MPB ruble data are used for 1928 and a 1928 price of 91.2 rubles per
1000 KWH is used to get 1933 ruble estimates.
3
1
45
268
111
039
540
35
3 130
140
815
9 30
11
1
2
210
700
400
870
820
(254)
15 000
(484)
(2 262)
4 000
-19-
Entries for the input column of the electr ic power sector are not readily
available from plan documents, so I have applied the estimates developed by
the Indian S ta t i s t ica l Inst i tute for India in 1959/60, as resorted by Richard
S. Eckaus and Kirit S. Parikh in Planning for Growth (1963), p. 60. The
principal flows involve coal, ra i l transport of coal, and the diagonal cell
(including losses).
The f i rs t FYP lacks financial detail for a transport and communications
flow-table row, but some dimensional evidence is available from other sources,
especially a s t a t i s t i c a l publication issued by the transport commissariat
setting forth the plan's transport aspects (though mainly for the "basic
variant"), Actual and expected revenues from freight and passenger service
by the several carriers are set forth in Table 5; the dominant role of the
railroads is apparent.
Table 5. Transport sector revenues, 1928 and 1933, by mode, in millionsof rubles
CarrierRailroads (27/28State steamships (1928)Soviet trade fleet (1928)Caspian steamships (1928)Sea ports (27/28)Waterways 27/28)
Total, a l l carriers
Railroads (32/33State steamships (1933)Soviet trade fleet (1933)Caspian steamships (1933)Sea ports (32/33)Waterways (32/33)
All transport
All transport under"optimal variant"
1928Total operating
revenue1721.2
102.928.428.9
9 . 0.7
1891.1
19332699.5
197.261.837.319.7
1.13016.6(a)
Passengerservice317.8
20.23.91.3. . .
343.2
407.027.6
6 .81.2
442.6
Freightand other
1403.482.724.527.6
9 .0.7
1547.9
2292.5169.655.036.119.7
2574.0
3460.3. 507 .7 2952 .6
(a) Sum of componen t s ; c a r d 1 i n s o u r c e g i v e s 3 0 1 4 . 0 .S o u r c e : Nar . komm. p u t e i ' s o o b . , Tsen . p l a n , u p r a v . , o t d e l s t a t . i k a r t . ,
P i a t i l e t n i i p l a n t r a n s p o r t a na. 1928/29 — 1932 /33 g . g . (Moscow, 1 9 2 9 ) ,c a r d s 1 , 4 9 , 6 2 , 70 , 6 9 , and 6 1 .
-20-
The introduction to the transport commissariat plan document explains that the
optimal variant of the plan "actually does not change the basic lines of the
basic variant but only strengthens i t s several parts," Some of the major
railroad bench marks are shown in their optimal-variant form, and in particu-
lar, the anticipated 1933 total operating revenue figure of 2,699.5 million
rubles is raised in the optimal variant to 3,096.6 million, an increase of
14.71 percent. I have therefore applied this expansion factor to the a l l -
transport targets to obtain "optical variant" totals .
For both the 1928 and the 1933 flow tables, these freight revenues
must be allocated to various receiving sectors along the transport and com-
munications row. The composition of Soviet freight traffic was such that
the great bulk, of these charges was paid by industry on i t s inputs. Without
seeking to compile data on ton-kilometers carried and 1926/27 rates charged
for each commodity group, I draw on NKPS estimates for the percent composi-
tion of ra i l freight traffic in 1927/2S and in the basic variant for 1932/33
to obtain rough measures of deliveries to producer goods industry, consumer-
goods industry, and urban households. The commodity groups covering coal
and coke, o i l , timber, ores, and iron-and-steel accounted for 48.22 of
1927/28 ton-kilometers; their share was expected to be 41.0% in 1932/33.
These shares are applied to al l -carr ier freight revenues to yield entries
for producer-goods industry outlays on the transportation of their inputs.
Movement of grain and salt similarly accounted for 15.7% and 14,1% of ra i l
ton-kilometers, and these shares are used to obtain an estimate of outlays
by consumer-goods industry. The 5.6% and 4.0% of ra i l ton-kilometers in-
volving firewood reflects a large scale flow mainly sent to urban households.
The remaining one-third of freight traffic is not allocated under this
- 2 1 -
procedure, but parts of i t vent to each of the three principal recipient
sectors, so fuller information might not change the breakdown drastically,
A 1927/23 ruble figure for ra i l revenues earned by railroad administrations
carrying traffic for the account of other railroads ("company freight" in
U.S. ra i l records), is taken from A. M. IAkobi, Zheleznye dorogi SSSR v
tsifrakh (1935), pp. 64-65, and raised in proportion to all freight revenue
for 1932/33. By crude assumption I assign 50% of the outlays on passenger
transport to urban households, 30% to rural households, and 20% to govern-
ment travellers, placing these amounts in the relevant cells of the transport
and communications row. The f i rs t FYP, I I , 2, p. 318, puts operating revenues
for the communications sector at 171.8 and 262.1 million rubles for 1928 and
1933 respectively; in the absence of any detailed evidence I assume that 40*
of these charges were paid by government, 20% by urban households, 10% by
rural households, and 15% each by producer goods and consumer goods Industry.
The trade— and-distribution sector 's contribution to economic activity,
being nonmaterial, was not much noticed in the f irst FYP , but i t s role can
be traced in the records. I ts "net product" appears in the national income
accounts summarized in FYP? , Vol. I, at p. 158, where i t is equated with the
"trade markup" (vyruchka) and placed at 2,825 million rubles for 1923 and
5,958 for 1933. Since the services were delivered primarily in connection
with manufactured goods, I have divided these amounts between peasants and
workers in proportion to the outlays that each population category was ex-
pected to make on manufactured goods, drawn from the data in FYP t Vol. I I ,
part 2, pp. 72-75.
Another category of nonmaterial output whose role deserves recognition
in our input-output tables is government services. Total ruble outlays
were estimated in the first FYP as follows (in millions of rubles):
1
1
2
1928
029.0607.6
90.2127.4100.0954.2
1933basic
variant
2 3391 080
2011 460
1275 207
targetsoptimalvariant
2 5701 258
2991 485
1475 758
EducationHealth careSocial insuranceAdministration and defenseCourts and security
Total
Source: FTP, Vol. I I , part 2, pp. 258 and 388-89.
These services went partly to the rural and urban population (education,
health care, social security), and partly to the state (administration, de-
fense, courts, and security). Services directly to the public can be divided
between rural and urban consumers on the basis of a table in MPB (p. 164)
allocating "free socio-cultural services" for 1928 in 1923 prices. I have
applied the same proportions to the 1933 target.
The fixed capital column in the final-demand quadrant of the 1928 and
1933 flow tables records four types of delivery. The first reflects the
growth of the fixed capital stock in livestock-products agriculture, where
herds of catt le, horses, camels, and bullocks, together with flocks of sheep
and pigs, and even hives of bees, all made up a large and growing category
of capital. The second and third involve flows of equipment from producer
goods and consumer goods industry to go into the buildings and structures
coming into use. The fourth reflects the gross output of the construction
sector, at least the completed portion of i t . Unfinished plant is delivered
to the inventory column. Gross output totals for 1928 and 1933 are stated
in FYP, Vol. I I , part 1, pp. 452-53, for "outlays on construction of buildings
and structures (excluding equipment) in millions of rubles at 1926/27 prices":
-23-
3,887.7 and 13 726.1 respectively. The MPB tabulations for 1928 give a
total in 1928 prices of 4 239.1, of which 561.8 is an increase in inventory,
i . e . , unfinished construction. This proportion of unfinished to total con-
struction is applied to both the 1928 and the 1933 FYP
figures to obtain my flow table entries. The herd growth figure for 1928
is from MPB, p. 191. Their estimate is somewhat lower than the 1013 shown
in KTs 29/30, p. 448, which is in the same format and context as the FYP
figures (Vol. I I , part 2, pp. 68-69) of 1181 and 1619 for 1929 and 1933.
My 1933 entry enlarges the MPB 1928 estimate by the ratio 1619/1013.
In this model the housing sector is conceived as delivering a flow of
services to urban and rural households, derived mainly from the very large
stock of fixed capital on hand in 1928. The first FYP discusses the housing
problem, concentrating on urban housing, but contains no estimates of the
gross output of this nonmaterial form of production. Fragmentary data on
urban household expenditures suggest outlays of 894 million rubles in 1928
and 1,541 million rubles in 1933. The 1928 outlay implies a rate of return
on the existing urban housing stock of 7.17%. Applying i t to the existing
and intended rural housing stock yields estimates of 800 million rubles in
1928 and 1020 million rubles in 1933 for rural outlays, essentially imputed
rent on owner-occupied peasant housing.Capital stocks
The stocks of fixed capital on hand as of October 1, 1927 and October 1,
1928, together with those anticipated for October 1, 1932, were recorded
with a fair amount of sectoral detail in our four basic documents. The FYP,
Vol. I I , part 2, pp. 66-69, gives annual estimates for 1928-1933 in 1925/26
prices. Estimates for Oct. 1, 1927 in 1925/26 prices are in KTs 28/29, op.
-24-
426-29. After the FYP was published, KTs 29/30 presented estimates in
1926/27 prices—differing from the figures for Oct. 1, 1928, in 1925/26
prices by less than one percent—but instead of restating the Oct. 1, 1932
figures proportionately enlarged, so that output and capital would both be
in 1926/27 prices, I have retained the estimates that underlay Gosplan
computations in the FYP? itself. The estimates for Jan. 1, 1928 in MPB in
1928 prices suggest no significant changes but add very useful evidence
on unfinished capital.
Table 6 presents the unadjusted data. In assigning these capital stocks
to the ten KAPROST sectors, several subdivisions and adjustments are necessary,
While livestock herds are mainly capital for the livestock products sector,
horses provide draft power for the field crops sector, to which the capitalised
part of irrigation and land improvement outlays is also assigned. I have
arbitrarily divided the entries for economic structures 50-50 between the two
sectors and assumed that 3/4 of the "dead inventory" (plows, wagons, and other
agricultural equipment) relates to field crops; 1/4 to livestock products.
In rearranging the entries for industry, one must first deduct the
workers' housing attached to this account, then extend the split between
Group A and Group 3 so that it covers all planned industry, other state in-
dustry, and cooperative and private industry. It is also necessary to
project the 1928 proportions back to the preceding year. Among the branches
of Group A industry, Soviet statisticians included the production of building
materials (bricks, cement, etc.), but no separate sector for construction,
with its associated fixed capital, was recognized. I have had to assume that
the capital/output ratio estimated for India in 1959-60 and reported by Eckaus
& Parikh in Planning for Growth, p. 67, provides a relevant proxy for the
-25-
Table 6. Stocks of total fixed capital at the start of 1928, 1929, and
1933, by sector, in millions of rubles at 1925/26 prices.
1 Oct. 27
753
7
118
11
11
21110
66
632502089857
250
047008258540248253541952038641359886399
200
1 Oct. 28
853
743
119111
11
21110
70
001734280893
940550298814165059813011653286701974074656274971833
154
1 Oct. 32
9651
1912522123318
231131312
109
928762424504
044981330739514420203948263530592830819047765940635
190
Agriculture:Livestock herdsEconomic structuresDead inventoryIrrigation & land imp.
Industry:State industry, incl. housing
PlannedGroup AGroup 3
OtherHousing
Coop. & privateAll industryElectrificationTransportationCommunicationsTrade-warehousingEducationHealthAdministrationCommunal economyUrban housing excl. industryRural housing
Whole economy
capital/output ratio that prevailed in the USSR in 1928 and 1933. Multi-
plying the flow-table gross outputs of the construction sector by .153 yields
completed capital estimates which are deducted from the Soviet figures for
producer goods industry. The electrification figure for 1927 must be raised
to include the generating stations placed in the communal account (assumed
to have grown during 1928 at the same rate as regional and village stations);
the 155 million rubles involved are deducted from the communal account.
These fixed capital stocks include the value of unfinished projects, not
-26-
separately recorded in FYP? or KTs volumes. The MPB group did, however,
estimate the volume of capital in process as of January 1, 1928, 1929, 1930,
1931, and 1932, for selected sectors; Che figures for the first two years
in 1923 prices are as follows:
Jan. 1, 1923 Jan. 1, 1929
Total Unfinished % unfin. Total Unfinished %unfi
Census industry 5 712.1 880.4 15.4129 6 436.0 1 234.5 19.181
Transport & comm. 10 750.4 130.5 1.2139 10 934.6 302.7 2.763
Urban housing 12 204.6 183.0 1.4994 12 424.6 243.5 2.000
Trade & distribution 425.9 42.2 9.9084 464.3 49.0 10.553
Social-use capital 5 454.1 144.7 2.6531 5 698.0 203.3 3.567
The percent-unfinished ratio for industry is here applied to electric power,
producer-goods industry, and construction; the housing ratio is assumed to
apply to rural as ve i l as urban housing. MPB 32 being s i l en t , I assumed
that 10% of the fixed capital in both agricultural sectors and in consumer-
goods industry was unfinished in both years. The adjusted figures for com-
pleted fixed capital at the beginning of 1928, 1929, and 1933 are displayed
in Table 7, along with unfinished capital stocks at the beginning of each
year.
Depreciation rates applicable to each sector 's fixed capital can be
derived from the data set forth in FYP?, Vol. I I , part 2, pp. 67-71 (optimal
variant) . I take the ratio of iznos in 1923/29 to fondy at the end of
1927/23 as a treasure of depreciation applicable Co old-technology capital .
-27-
and the similar ratio for 1932/33 as a measure of depreciation rates appli-
cable to new-technology capital, without attempting an incremental computa-
tion. The sectoral rates are as follows:
Table 7, Adjusted stocks of completed and unfinished capital at the startof 1928, 1929, and 1933, by sector, in millions of rubies at1925/26 prices.
Completed fixed capital Unfinished capital
Population and labor force
Population and labor force estimates for the 1928-1940 period can be
constructed on the foundation of the 1926 Soviet population census, published
in substantial detail by 1929. The potential labor force was necessarily
related to the population aged 16-59; annual changes would reflect the entrance
-28-
of 16-year-olds and the exit of 60-year-olds. The first FYP reviewed many
details of labor force composition, focusing especially on the urban labor
force earning wages and salaries, and projecting changes in the role of the
proletariat vis-a-vis the bourgeoisie. 3ut for the KAPROST model, which
allocates labor to producing sectors as part of each solution, we can confine
ourselves to a measure of the potential labor force available in each year,
along with parameters controlling its allocation and specifying the resulting
earned incomes.
We begin, therefore, with Table 8, which shows estimates of the number
of people (males plus females) reaching age 16 and reaching age 60 for each
year from 1928 through 1941. In the absence of direct intercensal evidence,
these projections come from reported one-year cohorts in the 1926 census,
neglecting the normal mortality that would have thinned the ranks of each age
group after 1926, and thus exaggerating somewhat the labor force increments
hypothetically available. The unadjusted series is taken from the 1926 census
as transcribed in Frank Lorimer, The Population of the Soviet Union (Geneva:
League of Nations, 1946), pp. 231-32. Adjustments are required to smooth out
the pronounced bulges occurring at ages 12, 45, 50, and 55 as people mis-
report their ages. Careful smoothing for the young males was carried out by
Godfrey Baldwin of the Foreign Demographic Analysis Division of the U.S. Dept.
of Commerce in his Estimates and Projections of the Population of the USSR
(series P-91, No. 23, March 1973), p. 10, and I have merely applied his
corrective ratios to the male-female total. The corrections for those aged
45, 50, and 55 in 1926 are cruder. The peak year is assigned a figure repre-
senting the arithmetic mean of the five years centered on it; 352 of the
excess is transferred to the years immediately preceding and following the
-29-
Table 8.
Year ofbirth
191213141516
191718192021
1922232425
Projectedby year,
Age16 in
192829302132
193334353637
1938394041
additions to1928-1941, in
Unadjusted
3 6183 6664 2372 6692 901
2 3223 3353 0133 0583 543
3 6544 5434 4154 527
and subtractions fromthousands
Adjusted
3 6583 7253 7003 3032 976
2 8032 7452 8753 0543 373
3 9204.5334 4154 527
of persons.
Year ofbirth
186869707172
187374757677
1878798081
the population
Unadjusted
699659890
1 548630
750919568
2 175746
1 147952
1 0941 969
aged 16-5
Ad justed
699757
1 118895859
8481 079940
1 1111 118
1 3071 0461 3151 338
peak year; and 152 of the excess is assigned to the adjacent years. Chart 1, p. 30
shows that the smoothed results do not suppress all variation but do display
a more likely pattern than the unadjusted figures. Starting with a base-
period figure for the population aged 16-59 on April 1, 1928 (as set forth
in FYP, Vol. II, part 2, p. 164), we can add those entering at age 16 and
subtract those reaching 60 to bring forward an annual series to 1941.
The first FYP itself contains estimates of the population aged 16-59
from April 1, 1928 through April 1, 1933, together with estimates for the
engaged labor force (FYP, Vol. II, Part 2, p. 164). They imply the annual
increments set forth below, appreciably smaller than those derived above in
Table 8. It will be seen that Gosplan expected the ratio of engaged labor
to population aged 16-59 to rise from 63.06% to 67.76% by 1933.
April 1
192829303132
Population.aged 16-59
82 40384 67087 01988 99990 423
Annualincrement
2 2672 3491 9801 4241 147
Derivedincrement
2 9592 9682 5822 4082 117
Engaged000
51 96754 18956 51053 63360 421
labor force% participation
63.0664.0064.9465.8366.32
1933 91 570 1 955 62 044 67,76
In order to util ize the projected series for the population aged 16-59
as extended up to 1941, but take account of the smaller increments expected
by Gosplan for 1928-1933, one can compare Gosplan's engaged labor force series
with my population series, obtaining a lover participation rate—rising from
63.06% in 1928 to 65.01% in 1933—and then assure that the average annual rise
of .39% in the participation rate would prevail over the 1934-1940 period.
The results are presented in Table 9.
Table 9. Hypothetical population aged 16-59 and engaged labor force, by year,1928-1941, in thousands of persons
April 1
Population aged 16-59annual total
increments number
Engaged labor forcetotal participation
number rate (%)
- 3 2 -
The scattered data on employment and wages in plan documents raise many
problems of interpretation and coverage but suffice to provide estimates that
seen adequate for present purposes. Sectoral data on annual wages and employ-
ment, actual for 1923 and intended for 1933, appear in FYP, Vol. I I , part 2,
especially on pages 206 and 208, but also on pp, 9, 13, 16-17, and 48. Besides
the hired labor force employed in state enterprises, other nonagricultural
labor earned incomes in private and cooperative enterprises or as individuals.
The massive and intr icate problem of estimating and projecting earnings and
employment In agriculture is very sketchily dealt with in the first FYP.
The wage and employment figures assembled below are firmest for producer
goods and consumer goods industry, for transport and communications, for trade
and distribution, and for government services. The electric power annual wage
is put at 150% of the 1928 censes-industry wage, on the basis of the relation
shown for 1929 in TsUNKhU, Narkhoz 19 32, p. 460; 192 8 employment Is taken from
Ibid. , p. 422. Employment planned for 1933 is derived from the intended out-
put and an assumed improvement in output-per-man of 110%, comparable to the
highest expected in other high priority branches (FYP?, Vol. I I , part 2, p. 192).
The construction sector used both hired and "other" labor; I have had to
assume that the annual wage for "other" labor was the same as that in ag r i cu l -
ture and forestry. In the absence of any attention to the labor employed in
operating urban residential faci l i t ies or their annual earnings, I have placed
in this sector the servants and day laborers appearing in Gosplan tabulations.
The total nonagricultural wage b i l l in 1928 in Table 10 comes to 9,367
rubles, not very different from the 9, 107 million rubles estimated for urban
incomes by Gosplan (PT?, Vol. I I , Part 2, pp, 72-73). Gosplan's estimates
for workers outlays in this balance of incomes and outlays are clearlv in
purchasers prices, and are incomplete regarding nonhired labor. If non-hired
-33-
Table 10. Annual wage, labor force, and wage bill, by sector, 1928 and 1933,in rubles at 1926/27 prices.
Sector
Agri., field cropsAgri., livestock productsElectric powerProducer goods industryConsumer goods industry
Transport & communicationsConstructionHousingTrade and distributionGovernment services
labor outlays were proportionate to those of hired labor, the 1928 worker con-
sumption total would be 8, 646 million. Deducting excise taxes of 1,396 million
yields a workers consumption estimate in producers prices of 7,250 million rubles,
not drastically different from my flow-table compilation of 6,769 million.
The tota l nonagricultural wage b i l l in 1933 in Table 10 comes to 18,138
million rubles, substantially above my flow-table estimate of 11,99 7 million
of workers consumption outlays in producer:prices. The indicated Gosplan figure
for outlays in purchasers prices is 14,145 million which, when excise taxes of
2,245 million are deducted leaves 11,900 million as an estimate with comparable
coverage.
Agricultural incomes are considerably less certain. The FYP estimates of
agricultural incomes and outlays (FYP, Vol. I I , oart 2, pp, 74-75), include
the following items: 1928 1933
246Gross saving 880
Less: Money credit 19 5
Statistical discrepancy 7
Net saving
Taxes
44
545
41723
440
940
1923
8 533
9 122
.9354
6 769
9 367
.7226
1933
13 729
15 109
.9087
11 997
18 138
..6614
- 3 4 -
I have therefore added 589 million rubles to my compiled flow-cable estimate
for 1928 peasant consumption to obtain an estimate for aggregate 1928 agricul-
tural incomes, and similarly added 1,330 million to the 1933 flow-table peasant
consumption total to get 1933 agricultural income. The indicated "propensities
to consume " are:
Peasants: Consumption
Income
C/Y
Workers Consumption
Income
C/Y
There is little basis in plan documents for subdividing aggregate agri-
cultural income between field crop activities and the production of livestock
products. The rural population earned incomes in many diverse activities and
in fact I have had to ignore the fact that a good deal of their income came
from seasonal jobs in construction or industry. The values of the two output
categories combine with the structure of interindustry flows to leave far less
room for value added in livestock products than in field crops. Thus if, in
the absence of any clues, I assume equal per capita earnings In each branch,
and follow the manpower assignment estimates in FYP , Vol. II, part 2, p. 9 ,
the 1928 flow table will show substantial losses in livestock products. Not
having firm support for any such finding I choose instead to divide the
slender nonwage value added equally between the two branches, thus precipitating
the numbers in lines 1 and 2 of Table 10 . The figures for number of people
in each branch are not intended to convey findings growing out of independent
evidence.
- 3 5 -
Foreign tirade
Though planners were concerned about Soviet imports and exports, plan
documents contain very l i t t l e s tat is t ical analysis, presumably because pro-
spective developments were so uncertain. Not much information is needed, how-
ever, for the KAPROST model. Exrports have been estimated as part of the final-
demand deliveries made by the two agricultural sectors and the two industrial
sectors. The imports that provide modest supplements to domestic production
can be reconstructed from scattered plan tabulations and checked in the
historical stat ist ics issued by the Ministry of Foreign Trade.
Following the approach laid down by Eckaus and Parikh in their model for
India, I distinguish between the imports that are required in the production
of certain sectors' gross outputs, and imports that are optional, in the sense
that if sufficient export earnings or external credits are available, further
imports can augment domestic production. Imports of both types are possible
for field-crop agriculture, livestock-products agriculture, producer-goods
industry, and consumer-goods industry; the other six sectors of the model
cannot directly augment their domestic output through imports.
Imports of manufactured goods appear in a sources-and-uses table in the
first FYP (Vol. I I , part 2, p. 44). under the coy t i t l e , "other supply." Their
place in relation to total "trade output" of manufactured goods is indicated
by the following:
Trade output of mfg. goods (millions)
"other supply"
Trade output plus other supply
Percent of "other" to sum
1929 was a year of great strain in foreign trade and i t will be seen that
1927/2813 652
477
14 129
3.376%
1928/2915 586
429
16 015
2.679%
1932/33optimaltarget
26 023
1 255
27 278
4.601%
-36-
imports fell below the 1923 level though domestic production increased. This
suggests that the 1929 ratio of imports to combined imports-plus-domestic
trade output indicates a minimum of required imports. Applying this ratio to
1928 shows that 378 of the 477 were required and 99 were optional. Applying
the same ratio to 1933 implies that 731 of the 1255 would be required, while
an increased share, namely 524, would be optional under plan intentions.
These estimated amounts may be subdivided between producer goods and consumer
goods on the basis of import data for 1928 in MPB, pp. 187 and 203, providing
shares which are assumed to prevail in 1933 as well. We then have the following
figures, in millions of rubles:
Required Optional Total
Producer goods
1928 Consumer goods
All mfg. goods
Producer goods
1933 Consumer goods
All mfg. goods
Relating these required imports to the flow-table gross outputs, including
interindustry deliveries, yields the following required-import ratios:
Required Imports Gross output Imports/output
1928 producer goods 320 14 676 .02180
1928 consumer goods 58 7 152 .00811
1933 producer goods 619 27 117 .02044
1933 consumer goods 112 13 405 .00836
320
58
378
619
112
731
84
15
99
444
80
524
404
73
477
1063
192
1255
5
5
17
574
259
108
6
6
21
535
402
253
6
6
20
556
349
098
-37-
A similar approach can be used to deduce required and optional imports
of agricultural commodities, though the base-period evidence has to come from
the 1929/30 Congrol Figures (KTs 29/30, pp. 566 and 424-25) and no clues are
available for 1933 plan intentions. A series for "trade output, including
imports, at producers prices" can be compared with another for "trade output
at current prices" (with no mention of imports, and unfortunately for agricul-
tural rather than fiscal years) to precipitate implied imports, as follows, in
millions of rubles:
1927/28 1928/29 1929/30
Including imports
Excluding imports
Gross agri. output, excluding herdgrowth, in current prices
Ratio of imports to gross output .01841 .006256 .01030
The 1928 figure of 315 compares well with a figure of 331 in Min. Vnesh.
Torg., Vneshnaia Torgovlia SSSR za 1928-1940 gg., p. 18. Of this total, some
273 million can be identified therein as involving the following major com-
modities:
Field crops: Cotton 154 Livestock products: wool 61
Vegetables & fruits 9 Large hides 39
Grain 8 Small hides __2
171 102
Applying these proportions to the total of 315 yields the following:
Required Optional Total
1928 field crops 67 130 197
1928 livestock products 40 78 118
Relating these required imports to my flow-table gross outputs produces ratios
of .00609 and .00649, which are assumed unchanged for 1933.
B.1
Appendix 3: Logic and Algebra of the Model
The KAPROST model exists in two forms. The first or structural form
is a familiar set of economic relationships derived from a complete, three-
quadrant input-output system, augmented by production and investment relation-
ships. This version is easy to understand, but contains a very large number of
variables and equations. In order to make the problem more manageable for
computer solution, a second, reduced form of the model was created. It is
derived from the first form by eliminating redundant variables and equations
through substitution. This version is more difficult to understand, but much
less costly to solve.
These aspects of KAPROST are described in three sections. The first
section explains the terminology and symbols used in the model. The second -
and third sections describe structural and reduced forms respectively.
1. Terminology and Symbols
Tables 1 through 4 present concise descriptions of the various terms,
variables and parameters used in the model. The input-output system upon which
KAPROST is based consists of ten productive sectors, listed in Table 1. Both
alphabetic abbreviations and numerical indices are used, the first because
they are easy to remember, the second because they are convenient in specifying
the model equations. The producing sectors may be further grouped as follows.
The agricultural sectors, field crops and livestock, are contrasted with all
other sectors in that our "peasant" and "worker" classification is defined in
of electric power, producer goods, consumer goods, transport and communications,
and, construction. Housing, trade and distribution, and government services
are the service sectors of the economy.
Table 2 relates the periods of the model to historic time. Our pre-
solution base year is 1923, corresponding to period 0. The model is defined
3.2
-28-
Table 1
Sector Names, Abbreviations and Indices
NAME Abbreviation Index
Agriculture-Field Crops
Agriculture-Livestock
Electric Power
Producer Goods
Consumer Goods
Transport and Communications
Construction
Housing
Trade and distribution
Government Services
Throughout the model coding, where an arbitrary sector is to be denoted, we
will use subscripts "i" and " j . "
AF
AL
EP
PG
CG
TC
CO
HO
TD
GS
1
2
3
4
5
6
7
8
9
10
.3
Period Number
0
1
2
3
4
5 .
6
7
8
-29-
Table 2
Time Periods
Years Covered
1928
1929 and 1930
1931 and 1932
1933 and 1934
1935 and 1936
1937 and 1938
1939 and 1940
1941 and 1942
1943 and 1944
Comment
Pre-Solution Base Year
Initial Solution Period
Model Solution Periods
Terminal Solution Period
Post-Solution Periods
The subscript "t" is used to denote an arbitrary period. The subscript value
of "T" denotes the terminal solution period when we wish to indicate that the
time horizon of the model need not be limited to 6 periods. "T+l" then
indicates the first post-solution period, etc.
3.4
in terms of eight periods of two years each, of which the first six are fully
solved for all economic activity. The last two are post-solution periods,
but are of concern to the model in that investment decisions made in the last
two solution periods will affect additions made to capital stock in these post-
solution periods. Note that while our planning horizon is limited to six
periods or twelve years, there is no theoretical reason why this could not be
extended.
The variables used in the model are listed in Table 3 along with a
brief description. Note that all variables have implicit time subscripts and
technology superscripts. Further, some variables require a subscript indi-
cating producing sector, and four of the variables are exogenously specified.
A variable not specified exogenously is solved for by the linear programming
algorithm.
Table 4 describes the parameter and coefficient names, and their
appropriate sub-and superscripting. All of the values for these items are
specified exogenously, though the manner of their derivation differs. Most
parameters are derived from Soviet economic data as described in appendices
A and C. Some, however, are more appropriately considered control parameters,
in that their value is reset by the modeler in order to alter the character-
istics of the solution. The peasant and worker average propensities to consume
(CX., A ) , the post-terminal growth parameters (^), the social discount rate
(r) and the fractions of peasant and worker consumption which must be satisfied
in fixed proportions (D ,j^ ) are control rather than economically derived
parameters.
2. Structural Form of the KAPROST Model
The structural form of the KAPROST model equations are given in Table 5,
and our discussion below refers to the equation numbers used there.
3.5
-30-
Table 3
Glossary of Variable Names
Name
A
CP
cw
D
E
F
G
H
J
K
Description
Foreign Trade Credits (+) or Debits (-)
Total Peasant Consumption
Total Worker Consumption
Amortization Expenditure (Depreciation)
Exports
Consumption
Government Expenditures
Deliveries to Inventory
Inter-Industry Uses
Capital Stock
KAPIPOT Value of the Objective Function
L Labor Force in Use (Required)
M Imports (Required)
N Deliveries to Fixed Capital
0C Optional Consumption0M Optional/Discretionary Imports
S Available Labor Force
X Gross Output
YP Peasant Income
YW Worker Income
Z Additions to Capital Stock
Varies bySector ! Period
exogenousVariable
X
X
X
X
X
X
X
X
X
X
X
X
X
x
X
X
X
X
X
X
XX
X
X
X
X
X
X
X
X
X
X
A variable which varies by sector will have its sector index as its firstsubscript; the time period index will always be the last subscript.Thus "X " is the gross output of sector "l" in period "t"
Note the inter-industry flow variable, J, requires two sector indices; thefirst for the producing sector, the second for the consuming sector.
3.6
-31-
Thus "J " is the delivery of output from industry "i" to industryijt
" j " in period "t."
A superscript indicates type of technology: "0" for old, "N" for new.
Name
a
-32-
Table 4
Glossary of Coefficients
Description
Inter-Industry Technical Coefficient
Peasant Average Propensity to Consume
Output-Capital Ratio
Worker Average Propensity to Consume
Depreciation Rate2
Varied byTechnology Sector
Post-Terminal Capital Stock Growth Parameter
Mastery Coefficient for Additions to CapitalStock
Output-Labor Ratio
Required Import Coefficienti
Structure of Worker Consumption
Structure of Investment Demand"3
Structure of Peasant Consumption
Social Discount RateFixed-proportion fraction of peasantconsumption
Deliveries to Inventory
Wage Rate
Fixed-proportion fraction of workerconsumption
Coefficients are all exogenously specified. A coefficient which varied bytechnology will have a superscript to indicate the type: "0" for old,"N" for new. Sector is indicated by a subscript. Thus: "XN" is theoutput to labor ratio for new technology production in sector "i".
Notes :(1) The inter-industry technical coefficients require two sector subscripts.
Thus lvaP^ ' is, for old technology, the deliveries required from sector"I" in order to produce one unit of output in sector " j . "
(2) The depreciation rate is the amount of each unit of capital which is"used up" in one year, hence is unavailable for future production..
3.8
-33-
Table 4 (cont'd.)
(3) Investment in capital follows a rigid, two-period (r year) schedule.
Thus "P " is the initial period deliveries to fixed capital from
sector "i" to eventually yield one unit of new capital in sectorVT
" j " , and "P* " is the second period deliveries.
(4) Wage rate varies by sector and by period. "W " is the wage paid
labor in sector "I" in period "t."
B.13
A linear programming model is an optimizing model, and so requires an
objective function. Equation 1.0 presents an objective function which contains
peasant and worker consumption in each year, and new technology capital stock
in the first post-terminal year. The initial weighting scheme sets all weights
for the year 1940 to 1, and raises the weight on consumption in earlier years
by an appropriate power of the social discount rate in order to compensate for
the passage of time. Clearly these basic weights are only a starting point for
analysis, and in the course of simulation the coefficients on ail items in the
objective were changed in order to alter the characteristic of the solution.
Note that the objective function captures the primary dilemma of development:
the time pattern of consumption, investment and capital formation by sector.
The distribution relationships are composed of the row identity of
input-output system (2.1) which limits the sum of the uses of any sector in any
period to the sum of gross output and imports of that sector in chat period.
Equation (2.2) defines interindustry demand as a result of the gross output of
each sector times the appropriate technical coefficient, for old and new
technology production. Equation (2.3) shows the dependence of deliveries to
fixed capital on the increments to capital stock in the next two periods.
Capital requires a two period (four year) gestation period in KAPROST, with
coefficients for the investment which must occur in each of the two periods in
order for capital stock to increase.
Exports and government spending are exogenously specified. Equation
(2.4) shows the three components of consumption: the fractions of peasant and
worker consumption which must be satisfied in fixed proportions and optional
consumption. Our Soviet data shows a fairly stable pattern of actual 1928 con-
sumption and planned consumption in the first five year plan. We interpret this
as representing consumer preference and require part of the consumption demand
of peasants and workers to be satisfied according to the fixed proportions of
3.14
these patterns. The rest of their consumption may be optionally filled from
any sector at the discretion of the solution algorithm.
Production requires some level of inventories, and in equation (2.5)
deliveries to inventory are given as a fixed fraction of output. Similarly in
equation (2.6) imports required for production are given as a fixed fraction of
output, and total imports by sector and period are the sum of these and optional
imports. Note that required inventory and import levels vary by the producing
technology. Equation (2.6) indicates that gross output is the sum of that
produced under old and new technology.
The capital accounting relationships record the effects on initial
levels of capital stock of depreciation and new increments. Equation (3.1) an
(3.2) are for old technology and new technology capital respectively, with the
only difference in treatment due to our limiting investment to new technology
production. Equations (3.3) calculates total depreciation.
Production is limited to a fixed multiple of capital stock as indicated
in equations (4.1) and (4.2). The complexity of the relation for new technology
production results from the requirement that increments to capital stock are not
fully productive in the year they are made. Rather, only the fraction k of
additions to capital, plus the undepreciated fraction of the previous period's
capital stock are available for production. This specification is intended to
represent the lower productivity which occurs while new capital is mastered by
the labor forced, or worked in to use.
KAPROST uses the output-labor ratios to determine the required labor
force for the output produced in each sector (equations (5.1) and (5.2), subject
to the constraint of total labor force (equation (5.3)). Note that one result
of this specification is that unemployment, a phenomenon never observed in the
Soviet Union, is possible. Rather than interpreting unrequired workers as
3.1
unemployed, it is probably more accurate to consider them as supernumeraries
in their place of work..
Income is a result of wages to required labor, as given in equations
(6.1) and (6.2). Note that wages do not vary by technology. Equations (6.3)
and (6.4) express, for peasants and workers, the requirement that total con-
sumption be at least a given fraction of total income. These fractions may be
viewed as average propensities to consume, or as inflation factors relating
nominal income to real wage in consumption goods. In equation (6.5) we require
total consumption to be satisfied in part by a fraction which must be delivered
in fixed proportions of output from the producing sectors, and in part by
optional consumption delivered from any sector.
Equation (7.1) requires that the foreign trade accounts for each year
be in balance or in surplus.
Finally, the terminal investment requirement (equation (8.1)) forces
deliveries to investment in the last solution period be high enough to allow
additions to capital stock in the second post-terminal period to exceed average
additions over the model solution period, plus some fraction of the final capital
stock.
Note that each equation in Table 5 is simply a template. They must be
expanded into separate equations either for each sector or for each time period
or for every combination of sector and time period in order to completely
specify the model.
3. Reduced Form of the KAPROST Model
The reduced fora of the KAPROST model is presented in Table 6. It is
obtained by substituting the equalities of the structural form into the in-
equalities so as to eliminate redundant variables. Here we indicate the sub-
stitutions made.
B.19
The objective function, equation (1.0) remains unchanged. The new
distribution relationship (2.1) is obtained by substituting the definitions
of each of the components of the input-output table row constraint into that
constraint, grouping common factors, and isolating all exogenous variables on
the right hand side. This last step normalizes the form of the problem for
the linear programming solution algorithm, and is done for all of the equations
below.
The capital accounting relationships (equations (3.1) and (3.2) repeat
the time progress of new technology capital as it appears in Table 5, making
explicit the exogenous setting of the initial period new technology capital stock,
and second period additions to new technology capital stock.
Accounting relationships for old technology capital stock may be disposed
of; no investment in old technology capital is permitted, the exogenously
specified initial value simply depreciates so that the stock for each year is
effectively exogenous. This is most evident in the productivity relationship
for old technology capital (equation (4.1)) where output from old technology
is limited by the remainder of the initial stock after the appropriate number
of years of depreciation. The relation for new technology in equation (4.2) ,
notes the effect of the mastering requirement for additions to capital. Note
that all variables in equations (4.2) appear on the left hand side as they are
all solved for by the model.
The labor supply constraint given in equation (5.1) is obtained by
substituting output divided by the output to labor ratio for labor. This sub-
stitution is also made in the income-consumption relationships (6.1) and (6.2)
to calculate income and then relate it to consumption by the average propensities
to consume. The requirement that total consumption be properly divided between
a part received in fixed proportions and an optional portion is normalized in
equation (6.3).
3.20
The balance of payments constraint in equation (7.1) shows the substi-
tution of the definition of required imports into the structural form in Table
5. The terminal investment constraint (equation (8.1)) has simply been
normalized.
As with the structural form equations in Table 5, the reduced form
equations in Table 6 are only templates. They must be expanded for all
appropriate sectors and time periods in order to yield a complete linear
programming problem.
C.I
Appendix C: Data as Modified for Use by the
KAPROST Model
The data estimates from Soviet and other sources presented in
Appendix A need to be reformulated in order to correspond with the KAPROST
model described in appendix B. These calculations, the motivating assumptions,
and the data used are described here. Note that this appendix presents the
data values which correspond exactly to the problem free solution. While most
of these values also underlie the historic simulations, the reader should be
aware of the distinction. The alternatives used for specific historic simula-
tions are described in the body of the report with the simulation results.
1, Exports
Soviet sources provide historic export levels for 1923, and those
forecast for 1933 by the first five year plan. The problem free solution is
assumed to provide these forecast amounts, and for years between 1928 and 1933,
and after 1933, amounts given by linear interpolation and extrapolation. Table 1
gives the yearly values so obtained. As the KAPROST time periods are two years
each, these values are summed appropriately to yield the exports for each of
the six solution periods of the model (Table 2). Note that only the agricultural,
(both field crops and livestock),producer goods and consumer goods sectors par-
ticipate in foreign trade.
2. Government Expenditures
The development of government expenditures is affected by our use of a
government services producing sector, which delivers output only to the govern-
ment. Our intention in doing this was to have all normal government expenditures
reflected in demand for the output of the government services sector, which
required interindustry deliveries from other producing sectors, and associated
that output with requirements for labor and capital stock. This provides,
C.2
Year
Table 1
Exports (millions of rubles)
Industry
AF AL PG CG
1928192919301931193219331934193519361937193819391940
7684.492.8
101.2109.6118126.4134.8143.2151.6160.0168.4176.8
230258.8287.6316.4345.2374402.8431.6460.4489.2518546.8575.6
263370.6478.2585.8693.4801908.6
1016.21123.81231.413391446.61554.2
50 -62748698110122134146158170182194
The 1928 values are actual exports; the 1933 values are those projectedby the first five year plan. All others are the result of linearinterpolation or extrapolation from the 1928 and 1933 figures.
AF: Agriculture, Field Crops
AL: Agriculture, Livestock
PG: Producer Goods
CG: Consumer Goods
C.3
Period
Table 2
Exports by KAPROST Period(millions of rubles)
AF
Industry
AL ?G CG
1929 and 1930
1931 and 1932
1933 and 1934
1935 and 1936
1937 and 1933
1939 and 1940
177.2
210.8
244.4
278.0
311.6
345.2
546.4
661.6
776.8
892.0
100 7.2
1122.4
843.3
1279.2
1709.6
2140.0
2570.4
3000.3
136
184
232
280
328
376
Values reflect two year totals from Table 1.
AT: Agriculture, Field Crops
AL: Agriculture, Livestock
PG: Producer Goods
CG: Consumer Goods
These numbers provide values for the variable "E" in KAPROST for sectors1 , 2, 4 and 5. Exports of all other sectors are zero.
C.4
again for normal government activity, a neat way of combining deliveries from
other sectors, labor force and productive capital stock, rather than associate
these with a final demand category directly. For other than normal scenarios,
e.g. rearmament, government demand for producers goods can be specified directly,
as tanks, planes, ships, etc., are produced by this sector yet do not enter
into capital stock, but are considered to be consumed. However, the labor force,
barracks, and common needs of soldiers can be provided for by increasing
government expenditure on the output of government services, as this increases
demands on all these items in proportion, plus forcing additional wage payments
and consumption demands to round out the non-heavy goods requirements of the
military.
For the problem free solution we again took actual 192S and projected
1933 government expenditures and obtained values for intervening and succeeding
years by linear interpolation and extrapolation. These were summed over two
year intervals to obtain the G10 variable in KAPROST. Direct government
expenditures on the output of all other sectors is zero. These results are
presented in table 3.
3. Labor Supply
Yearly labor force statistics must be converted to values appropriate
to KAPROST's two-year periods. This is done by taking the average labor force
over the two years, and is presented in table 4.
4. Initial Levels of Capital Stock
The basic data for capital stock consist of actual 1928 and 1929 levels,
plus an indication of additions to capital stock based on projects in progress
in 1928. These last provide direct estimates for Zl» additions to new tech-
nology capital in the second period, as given in table 5. We must therefore
c. 5
disentangle the 1923 and 1929 stock figures to provide estimates of initial
period capital stock of both technologies.
Our basic assumption is that all capital stock in 1928 is of the old
technology, and no additions are made to it so that the only process affecting
it is depreciation. Hence for each sector, the initial period stock of old
technology capital is equal to the 1923 stock of capital minus depreciation.
The initial period stock of new technology capital is the difference between
the total 1929 stock and the calculated initial stock cf old technology capital
These calculations are reproduced in table 6.
Year
Table 3
Government "Expenditures on Government Services(in mil l ions of rubles)
G10by Year Period G10
1923192919 301931193219331934193519361937193319 391940
2954.03554.84155.64756.25357.25958,06558.87159.67760.48361.28962.09562.3
10163.6
1929
1931
1933
1935
1937
19 39
and
and
and
and
and
and
1930
1932
1934
1936
1938
1940
7710.4
10113.6
12516.8
14920.0
17323.2
19726.4
G10 is the government expenditure on the output of sector 10,government services. Direct government expenditure on theoutput of all other sectors is zero (though indirectly, thproduction of G10 requires output from several sectors).
C6
Year
Table 4
Labor Supply (in thousands)
LaborSupply Period
Labor
1923192919 3019 31193219331934193519361937193319 391940
51967541895651058633604216204463694651706683768524704367260675339
1929
1931
1933
1935
1937
1939
and
and
and
and
and
and
1930
1932
1934
1936
1938
1940
55350
59527
62869
66004
69430
73998
The period labor supply is the average of the yearly labor suppliesin that period.
The last column data provide values for the variable S in KAPROST.
Table 5
Additions to New Technology Capital Stockin Period 2
Sector
(millions of rubles)
N
AFALEPPGCGTCCOHOTDGS
3151140194
1308697"33115947974
213
.7
Table 6
Initial Period Stocks of Capital(millions of rubles)
Sector
(1)
1928CapitalStock
(2)
DepreciationRate
(3)InitialPeriodOldTechnologyCapitalStock
(4)
1929CapitalStock
(5)InitialPeriodNewTechnologyCapitalStock
AFALEPPGCGTCCOHOTDGS
8224759460030172618
11366595
22982487
5680
.087
.0768
.0524
.0517
.0509
.0358
.0994
.039
.0485
.0284
7509701156928612485
10959536
220864635519
84887965817
33852937
11608667
23490627
5765
979954248524452649131
1404164246
Column 1 and 4 are historic values
Column 3 = (1-Column 2) * Column 1
Column 5 = Column 4 - Column 3
Column 3 provides the exogenous values for K_,
Column 5 provides the exogenous values for K?
,1
C.3
5. Depreciation Coefficients
The one-year depreciation rates obtained from Soviet data need to be
converted to the two-year model periods. As the depreciation rate gives the
proportion of each unit of capital used up in one year, one minus this rate
is the proportion remaining after one year. The square of the proportion
remaining after one year is the proportion remaining after two, and the
difference between this and one the two-year depreciation rate. These calcu-
lations are repeated in Table 7.
6. Fixed Proportion Structure of Consumption
The sector proportions in which peasants and workers would choose to
consume, given the opportunity, are based on actual 1923 consumption and planned
1933 consumption. As these proportions did not change significantly over the
five years, the consumption-weighted averages of the two observations were used.
Table 8 presents these calculations.
7. Wages
The actual 1928 wages and planned 1933 wages by sector are used to
linearly interpolate and extrapolate values for the remaining years (Table 9).
These values are summed over two-year intervals to provide the sector wages
for KAPROST (Table 10).
8. Investment Structure
The time pattern of deliveries from capital-goods producing sectors
to investment demand in order to eventually make a new unit of capital available
to a given sector is a combination of two sets of coefficients. Table 11 lists,
for each sector, the proportions of Livestock, Producers Goods and Construction
which go into one ruble of capital stock. It is then assumed that capital
is produced over two, two-year periods, with .2398 of the deliveries occurring
C.9
Table 7
Depreciations Rats Calculations
Sector
Old Technology:AFALEPPGCGTCCOHOTDGS
New Technology:AFALEPPGCGTCCOHOTDGS
(1)One Year
DepreciationRate
.0870
.0768
.0524
.0517
.0509
.0358
.0994
.0390
.0485
.0284
.0753
.0764
.0517
.0447
.0460
.0359
.0859
.0360
.0509
.0258
(2)RemainderAfter One
Year
.9130
.9232
.9476
.9483
.9491
.9642
.9006
.9610
.9515
.9716
.9247
.9236
.9483
.9553
.9540
.9641
.9141
.9640
.9491
.9742
(3)RemainderAfter Two
Years
.83357
.85230
.89795
.89927
.90079
.92968
.81108
.92352
.90535
.94401
.85507
.85304
.89927
.91260
.91012
.92949
.83558
.92930
.90079
.94907
(4)Two Year
DepreciationRate
.16643
.14770
.10205
.10073
.09921
.07032
.18892
.07648
.09465
.05599
.14493
.14696
.10073
.08740
.08988
.07051
.16442
.07070
.09921
.05093
Note:
( 1 ) : from Appendix A(2) = 1.0 - (1)(3) - ( 2 ) 2
(4) - 1.0 - (3)
The data in (4) provides the values for d and <f in KAPROST,
C.10
Table 8
Fixed Consumption Structure
(1)1923
Consumption
(2)1923
Shares
(3)1933
Consumption
(4)1933
Shares
(5)Total
Consumption
(6)AverageShares
Peasants
AFALEP?G.CGTCCOHOTDGS
Total
14002995—34
2003120—8001181—
SS33
.16407
.35099—
.00398
.23474
.01406—
.09375
.13840—
1.00000
22934752—56
3284159—10202165—
13729
.16702
.34613—
.00408
.23920
.01158—
.07430
.15770—
1.00000
36937747—90
5287279—18203346—
22262
.16589
.34799—
.00404
.23749
.01253—
.08175
.15030—
1.00000
AFALEPPGCGTCCOHOTDGS
tal
6361051—59
2149286—8941644—
6769
.10134
.15527—
.00872
.31748
.04225—
.13207
.24287—
1.00000
11191668—93
3414369—15413793—
1199 7
.09327
.13903—
.00775
.28457
.0307.6—
.12345
.31616—
1.00000
18052719—152
5563655—24355437—
13766
.09618
.14489—
.00810
.29644
.03490—
.12976
.29973——
1.00000
Note:(1)(2)(3)(4)(5)(6)
from Soviet 1928 data(1) divided by column totalfrom First Five Year Plan(3) divided by column total(1) + (3)
(5) divided by column total.
Ail calculations performed separately for Peasants and Workersyields values for * and w in KAPROST.
Column 6
flows in mil l ions of rubles.
C.11
Table 9
Wages
Note: 1928 values are actual wages by sector.
1933 values are planned wages by sector
All other years are obtained by linear interpolation or extrapolation
All numbers in rubles.
C.12
Sector
AF
AL
EP
?G
CG
TC
CO
HO
TD
GS
Note:
(1)1929-1930
A98.8
498.8
2888.8
1882.6
1882.6
1862.2
1658.2
830.2
1655
1620.2
Two-year sums
(2)1931-1932
569.2
569.2
3271.2
2163.4
2163.4
2123.8
2127.8
915.8
1931
1897.8
Table 10
(3)1933-1934
- 639.6
639.6
3653.6 .
2444.2
2444.2
2385.4
2597.4
1001.4
2207
2175.4
from values in Table 9.
(4)1935-1936
710
710
4036
2725
2725
2647
3067
1087
2483
2453.
(5)1937-1938
780.4
780.4
4418.4
3005.8
3005.8
2908.6
3536.6
1172.6
2759
2730.6
(6)1939-1940
850.8
850.8
4800.8
3286.6
3286.6
3170.2
4006.2
1258.2
3035
3008.2
All values in rubles.
C.13
in the first of the two periods, and .7602 in the second. Table 12 shows the
p1 and p2 matrices which result from applying the time factors to the propor-
tional structure given in Table 11.
9. Parameters Which Vary by Technology:
Interindustry Technical Coefficients, Required Import Ratios,
Inventory Ratios, Output-Labor and Output-Capital Ratios.
Production by the two technologies available in KAPROST is completely
described by the interindustry technical coefficients, required import ratios,
inventory ratios, output-labor ratios and output-capital ratios. These numbers
are calculated incrementally from actual 1928 flow values and planned 1933 flow
values. That is, we assume that all 1928 activity is due solely to old tech-
nology production. After 1928 all investment is in new technology, so old
technology capital only depreciates and never increases. Hence, by subtracting
depreciated 1928 activity from planned 1933 activity we have as the remainder
flows attributable solely to new technology, allowing us to calculate parameter
values for new technology. Each parameter is described in turn.
a) Interindustry Technical Coefficients
Table (13a) lists the interindustry flows for 1928, and (13b) the old
technology coefficients obtained by dividing entries in each column by the
gross value of output of that sector. If we use the old technology one-year
depreciation coefficients for each sector from Table 7, and apply them over
the five years (1928 to 1933) to the 1928 gross value of output we obtain the
1933 flows which can be attributed to old technology (Table 13c). Table 13d
gives the planned, 1933 interindustry flows. Table 13e gives the 1933 new
technology flows obtained by subtracting table 13c from 13d. The calculation
of the new technology coefficients is performed by dividing each column of .
table 13e by the gross value of output of that column's sector. This result
C.14
Table 11
Investment Structure of Capital
SectorReceivingCapital
AF
AL
EP
PG
CG
TC
CO
HO
TD
GS
Capital Goods
AL
.1834
.6562
---
Producing Sector
PG
.5645
.1186
.6676
.4987
.4987
.4477
.7390
.3518
.4786
.1052
CO
.2501
.2252
.3324
.5013
.5013
.5523
.2610
.6842
.5214
.8948
Note: Each entry indicates the ruble value of deliveriesfrom a capital goods producing sector to investmentdemand to produce one ruble of capital in the sectorreceiving new capital.
C.15
SectorReceivingCapital
Table 12
Tine Structure of Capital Production
(1)First Period
(2)Second Period
Capital Producing Sector
AL PG CO
Capital Producing Sector
AL PG CO
AF
AL
EP
PG
CG
TC
CO
HO
TD
GS
.0445 .1354
.1574 .0284
.1601
.1196
.1196
.1074
---- .1772
.0757
.1148
.0252
.0600
.0540
.0797
.1202
.1202
.1324
.0626
.1641
.1250
.2146
.1409 .4291
.4988 .0902
.5075
.3791
.3791
.3403
.5618
.2401
.3638
.0800
.1901
.1712
.2527
.3811
.3811
.4199
.1984
.5201
.3964
.6802
Note: (1) = Table 11 x 0.2398
(2) - Table 11 x 0.7602
Ruble value of deliveries to investment to produce oneunit of capital available at the beginning of the. third period,
C.16
Table 13a
1928 Interindustry Flows
Receiving Sectors
ProducingSector
AF
AL
EP
PG
CG
TC
CO
HO
TD
GS
AF
154.1
350
—
145
191
—
—
—
—
—
AL
2853 '
125
5
96
133
—
—
—
—
—
EP
—
—
63
86
—
64
—
—
—
—
?G
424
—
.177
."5346
-746
981
—
—
—
—
CG
2944
510
108
3777
714
358
—
—
—
—
TC
97
—
17
650
152
116
—
—
—
CO
718
—
—
2119
11
—
—
—
—
- —
KO
—
—
—
142
—
—
—
—
—
—
TD
—
—
—
186
46
—
—
—
—
—
GS
57
40
196
100
672
137
—
—
—
—
InterindustryTotal 2227 3212 213 7674 8411 1032 2848 142 232 1202
Gross Valueof
Output 11002 6172 566 14674 7152 2062 3888 1694 2825 2954
Note: All values in millions of rubles
C.17
Table 13b
Old Technology Coefficients
Receiving Sectors
ProducingSectors
AF
AL
EP
PG
CG
TC
CO
HO
TD
GS
AF
.14007
.03181.
—
.01318
.01736
—
—
—
—
AL
.46225
.02025
.00081
.01555
.02155
—
—
—
—
EP
—
—
.11131
.15194
—
.11307
—
—
—
PG
.02889
—
.01206
.36432
.05084
.06685
—
—
—
.....
CG
.41163
.07131
.01510
.52810
.09983
.05006
—
—
—
TC
.04704
— ,
.00824
.31523
.07371
.05626
—
—
—
CO HO
.18467
—
—
.54501 .08383
.00283
—
—
—
—
__ __
TD
—
—
—
.06584
.01628
—
—
—
—
__
GS
.01930
.01354
.06635
.03385
.22749
.04638
—
—
—
Note: Obtained from table 13a by dividing each column by its gross value ofoutput.
0Provides a values in KAPROST.
. 10
Table 13c
Old Technology Flows in 1933
Receiving Sectors
Note: Obtained by depreciating table 13a overfive years. Rounded to nearest unit.Values in millions of rubles.
C.19
ProducingSectors
AF
AL
EP
?G
CG
TC
AF
2430
457
—
401
529
AL
4130
202
39
266
368
1933
EP
—
—
269
343
—
254
Table 13d
Interindustry F
Receiving
?G
626
—
1038
8788
1276
1576
lows
Sectors
CG
4747
670
330
6461
1222
535
TC
144
—
58
1547
361
193
CO
1059
—
—
5438
24
HO
—
—
—
492
—
__
TD
—
—
—
308
77
__
GS
90
62
545
326
1076
193
CO
HO
TD
GS
InterindustryTotal 3817 5005 866 13304 13965 2303 5621 492 385 2292
Gross Valueof
Output 17065 . 9577 2279 27117 13405 3279 13726 2561 5958 5958
Note: Values in millions of rubles.
C.20
Table 13e
New Technology Flows in 19 33
Receiving Sectors
InterindustryTotal 2404 2851 703 7419 7487 1442 3935 376 204 1250
Gross Value
of Output . 10086 5438 . 1847 15864 7897 1561 11423 1173 3755 3400
Note: (13e) = (13d) - (13c)
Values in millions of rubles.
C.21
is presented as table 13f.
b) Required Import Ratios
Table 14 lists the incremental calculations Co obtain the required
import ratios for old and new technology production.
c) Inventory Ratiosv
Table 15 lists the incremental calculations to obtain the inventory
ratios for old and new technology production.
d) Output-Labor Rat ios
The old technology output-labor ratios are obtained by dividing 1923
gross values of output by sector labor force. This is multiplied by two to obtain
productivity over the two-year KAPROST period. 1933 old technology labor force
equals 1933 old technology gross value of output multiplied by the one-year old
technology output-labor ratios. This is subtracted from total 1933 labor force.
We use this as the divisor for 1933 new technology gross value of output to
obtain new technology output-labor ratios. Again, multiplication by two adapts
the data to the two-year period of KAPROST. See Table 16.
e) Output-Capital Ratios
Table 17 shows the calculation of old and new technology output-capital
ratios. As this exactly parallels the calculation of the output-labor ratios
discussed above, no further concent is warranted.
C.22
ProducingSectors
AF .:
AL
EP
PG
CG
TC
CO
HO
TD
GS
AF
L4398
02330
03064
04043
—
—
—
—
AL
.40769
.02174
.00655
.03707
.05127
—
—
—
—
New
EP
—
.11963
.15018
—
.11108
—
—
—
Table
Technology
Receiving
PG
.01897 .
.05688 .
.29552 .
.04437 .
.05192 .
—
—
—
13f
Coefficients
Sectors
CG
31402
03510
03125
44983
08511
03284
—
—
—
TC
.04050
—
.02807
.64418
.15013
.06171
—
—
—
CO
.05547
—
---
.28737
.00153
—
—
—
—
HO TD
—
—
—
.32034 .04338
.01095
—
—
—
—
GS
.01194
.00806
.11037
.07041
.14531
.02188
—
—
—
Note: Obtained from table 13e by dividing each column by its gross value of output.N
Provides a values in KAPROST.
C.23
Table 14
• Required Import Ratios(1) - (2) (3)' (4)
Old Technology 1933Old Technology 1933
Required RequiredImports Importsctor
AF
AL
EP
?G
CG
TC
CO
HO
TD
GS
1928RequiredImports
67
40
—
320
58
—
—
—
—
. RequirImportRatio.
.00609
.00648
—
.02181
.00811
—
—
—
—
--
(5)1933
New TechnologyRequiredImports
(6)New Technology
RequiredImportRatio
42.5
26.8
245.4
44.7
67
40
619
112
24.5
13.2
373.6
67.3
.00243
.00243
.02355
.00852
(1) in millions of rubles0
(2) = (1) ÷ 1928 Gross Value of Output (Table 13a) , KAPROST m .
(3) = (2) x 1933 Old Technology Gross Value of Output (Table 13c) in millions of rubles
(4) in millions of rubles
(5) = (4) - (3) in millions of rubles.N
(6) = (5) ÷ 1933 New Technology Gross Value of Output (Table 13e) , KAPROST m .
C.24
Table 15
Inventory Ratios
(1) (2) (3) (4) (5) (6)Old 1933 Old 1933 New New
1928 Technology Technology 1933 Technology TechnologyDeliveries to Inventory Deliveries to Deliveries to Deliveries to Inventory
Sector Inventor' Ratios Inventory Inventory Inventory Ratios
AF
AL
EP
?G
CG
TC
CO
HO
TC
GS
206
—
—
30
234
—
515
—
—
.01872
—
—
.00545
.03272
—
.13246
—
—
__
130.6
—
—
61.3
180.2
—
305.1
—
—
__
309
—
—
350
1564
—
1819
—
—
178.3
—
—
288.7
1383.8
—
1513.9
—
—
.01768
—
—
.01820
.17523
—
.13253
—
—
(1) in millions of rubles0
(2) = (1) ÷ 1928 Gross Value of Output (Table 13a) , KAPROST s
(3) = (2) x 1933 Old Technology Gross Value of Output (Table 13c) in millions of rubles.
(4) in millions of rubles
(5) = (4) - (3) in millions of rubles
(6) = (5) ÷ 1933 New Technology Gross Value of Output (Table 13e) in millions of rubles,KAPROST sN
C.24
Table 16
Output-Labor Ratios
sect0
r
A?
AL
EP
?G
CG
TC
CO
HO
TD
GS
(1)
1928LaborForce
34770
3810
19
1767
3894
1410
1255
754
1065
2215
(2)One-Year
OldTechnology
Output-LaborRatio
.31642
1.61995
29.78947
8.30447
1.83667
1.46241
3.09801
2.24668
2.65258
1.33363
(3)Two-Year
OldTechnologyOutput-LaborRatio
.63284
3.23990
59.57895
16.60894
3.67334
2.92482
6.19602
4.49337
5.30516
2.66727
(4)
1933Old
TechnologyLaborForce
22058
2555
15
1355
2999
1175
744
618
831
1918
(5)
1933LaborForce
41310
4700
36
2754
4767
1545
209 7
868
1357
2754
(6)
1933New
TechnologyLaborForce
19252
2145
22
1399
1768
370
1354
250
526
836
(7)One—Year
NewTechnology
Output-LaborRatio
.52386
2.53532
85.95789
11.34034
4.46612
4.21836
8.43960
4.69041
7.13280
4.06658
(8)Two-Year
New;
TechnologyOutput-Labor'Ratio
1.04771
5.07063
171.91578
22.68068
8.93225
8.43672
16.37920
9.38082
14.26560
8.13315
(1) in thousands
(2) = 1928 Old Technology Gross Value of Output (Table 13a)* (1).
(3) 2.0 x (2) , KAPROST Xc
(4) = (2) x 1933 Old Technology Gross Value of Output
(Table 13c) in thousands
(5) in thousands
(6) = (5) - (4)
(7) - 1933 New Technology Gross Value of Output (TAble 13e)
(8) - 2.0 x (7) , KAPROST \n
* (6)
C.25
Table 17
Output-Capital Ratios
sector
AF
AL
EP
PG
CG
TC
CO
HO
TD
GS
(1)
1928Capital
Stock
8224
7594
600
3017
2618
11366
595
22982
487
5680
(2)One-Year
OldTechnologyOutput-CapitalRatio
1.338
.813
.94333
4.86377
2.73186
.18142
6.53445
.07371
5.80082
.52007
(3)Two-Year
OldTechnologyOutput-CapitalRatio
2.676
1.625
1.88666
9.72754
5.46371
.36284
13.06891
.14742
11.60164
1.04014
(4) (5)
1933Old
Technology 1933 .Capital CapitalStock Stock
5217
5093
458
2314
2016
9472
353
18837
380
2558
11600
10151
3191
9754
4867
18273
2100
28507
2318
10088
(6)
. 1933New
TechnologyCapitalStock
6383
5058
2733
7440
2851
8801
1748
9670
19 38
7530
(7)One-Year
NewTechnologyOutput-CapitalRatio
1.580
1.075
.67563
2.13217
2.76991
.17732
6.53677
.12125
1,93746
.45157
(8)Two-Year
NewTechnologyOutput-Capital
Ratio
3.160
2.150
1.35126
4.26434
5.53981
.35464
13.07353
.24250
3.87492
.90313
D.1
Appendix D: Programming and Computer Solution of the Model
The model described in Appendix 3 is an optimization model with linear
objective function and linear constraints, that is, it is a linear programming
model. We chose to use the IBM Mathematical Programming System (MPS) to
solve the model, both for reasons of availability and because it is one of the
most common linear programming packages in use. Our model was solved success-
fully by MPS on on IBM 360 Model 44 with 256 K of main storage. For a
detailed description of MPS please refer to the IBM publications listed at the
end of this appendix.
Use of MPS requires that all of the linear relations be formatted in
the MPS input language. A glance at the reduced form of the model (Appendix B,
Table 6) shows that the form of most relations is very regular, with entries
varying by industrial sector, time period or both. Further, even with the
economies in the reduced form description, the model still has over 300 con-
straints and over 600 variables, which would require a considerable amount of
keypunching to produce input acceptable to MPS. If this task were performed
only once, the direct keypunch approach would be feasible. However, our research
task required that we be able to vary parameter values quite often, and re-keying
changes would have been intolerable.
The regularity of the model, therefore, proved to be a blessing in
disguise, because it permitted us to write computer program which would accept
the parameters of the model as data, and then automatically generate the
appropriate input to MPS. This preliminary processing program, or PREL, was
written in the PL/I programming language, a choice dictated by the need for
character manipulation capabilities. PREL is highly idiosyncratic in that it
produces MPS input for only one model, KAPR0ST. For that reason we do not
describe it in detail (though a copy can be provided on request). However,
D.2
the use of this technique may be of interest to other researchers as a signifi-
cant time and labor saving device. For example, to raise capital productivity
for all sectors would require re-punching 120 cards, if the M?S input were
manipulated directly. Using PREL this can be done by altering two cards!
The solution obtained by MPS can be printed only in a rather crude
form: a list of variable names and data values. The economic theory behind
our work allows these solution values to be organized in the much more detailed
manner of input-output tables with inter-industry, final demand and value added
sectors. We felt it would be very useful in understanding the solution if the
data were displayed in this form. A post solution program, POST, was written
to take the output from MPS and reformat it in an economically meaningful way.
POST, in addition, explodes the solution variable values, which correspond to
the reduced form of the model, back to the complete set of variables used in
the structural form of the model (Appendix 3, Table 5). Like PREL, POST can
only be used for our particular model, but the idea is valuable. POST was also
written in the PL/I language and a copy can be provided upon request.
Figure 1 presents a flow diagram of our computer solution process.
The parameter values, described in Appendix A, are keypunched and the cards used
as input to PREL. PREL produces an MPS-formulated Model Description which is
accepted by MPS, in addition to a card-deck of MPS control statements. As some
need was felt for retaining different solutions in computer-readable form—
one, in particular, being the savings which obtain from starting the solution
process from a previous set of solution values rather than from scratch—a
magnetic tape serves as a long-term storage and retrieval medium. MPS solves
the model and produces a solution file, which is transformed by the POST program
into economically meaningful reports. POST also requires the model parameters
to produce the reports.